SEARCH

SEARCH BY CITATION

Keywords:

  • amphibians and reptiles;
  • life history evolution;
  • quantitative genetics

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Human activities are changing habitats and climates and causing species’ ranges to shift. Range expansion brings into play a set of powerful evolutionary forces at the expanding range edge that act to increase dispersal rates. One likely consequence of these forces is accelerating rates of range advance because of evolved increases in dispersal on the range edge. In northern Australia, cane toads have increased their rate of spread fivefold in the last 70 years. Our breeding trials with toads from populations spanning the species’ invasion history in Australia suggest a genetic basis to dispersal rates and interpopulation genetic variation in such rates. Toads whose parents were from the expanding range front dispersed faster than toads whose parents were from the core of the range. This difference reflects patterns found in their field-collected mothers and fathers and points to heritable variance in the traits that have accelerated the toads’ rate of invasion across tropical Australia over recent decades. Taken together with demonstrated spatial assortment by dispersal ability occurring on the expanding front, these results point firmly to ongoing evolution as a driving force in the accelerated expansion of toads across northern Australia.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

The process of range advance can have a profound impact on the evolution of vanguard populations (Excoffier & Ray, 2008; Phillips et al., 2010). Not only are vanguard populations typically at low density, but as the range advances, individuals on the advancing front are spatially assorted by dispersal ability: only the best dispersers in the population make it to the expanding front each generation. In theory, this ‘spatial selection’– spatial assortment, in conjunction with density-dependent population growth – can lead to strong directional selection on dispersal ability on the invading front and the evolution of phenotypes that are highly dispersive despite strong trade-offs between dispersal and survival (Travis & Dytham, 2002; Hughes et al., 2007; Phillips et al., 2008).

Rapid evolution of increased dispersal ability on an expanding front will have clear impacts on the rate at which the population spreads. Given that spatial selection is the outcome of a universal process (spatial assortment by dispersal ability on the invading front) and release from a common form of population growth (density dependence, Brook & Bradshaw, 2006), dispersal ability most often will be driven upwards on expanding fronts – thus causing the population to spread in an accelerating fashion (Holt et al., 2006). Such accelerating spread is a common feature of biological invasions (which often exhibit a characteristic lag phase before a rapid increase in range; Hengeveld, 1989) and may also explain a venerable biogeographical paradox – Reid’s paradox – which observes that current dispersal ability is insufficient to account for the rapid spread of trees in the northern hemisphere following the retreat of the last ice age (Clark, 1998; Phillips et al., 2008). Spatial selection helps explain the ubiquity of accelerated spread, and hence, to predict the likely rates of population shift in response to contemporary climate change (Thomas et al., 2001; Parmesan & Yohe, 2003; Mustin et al., 2009).

Despite the potential importance of spatial selection, few thorough empirical tests of the theory exist. Perhaps the first study to recognize the possibility of evolved shifts in dispersal on range edges was that of Cwynar & MacDonald (1987), who found seeds with a more dispersive phenotype occurring on the expanding range edge of lodgepole pines. Much later, in a landmark study, Simmons & Thomas (2004) showed that populations of crickets from an expanding range edge in Britain had higher proportions of long-winged individuals, that large wings were associated with longer flights in wind tunnels and that wing morphology had a genetic basis. Other empirical work supporting spatial selection comes from speckled wood butterflies, which are expanding their range in Britain. Speckled wood butterflies from the expanding range edge exhibit larger wings and thoraxes than their conspecifics from the range core (Hughes et al., 2007). Together, these studies represent the bulk of empirical support for spatial selection theory (although see also Léotard et al., 2009), although in all cases, they lack measures of actual dispersal in the field and were not linked to accelerating range advance.

The last line of support for spatial selection comes from our work on invasive cane toads in northern Australia. This work has documented behavioural and morphological shifts associated with increased dispersal rates in vanguard populations of toads (Phillips et al., 2006, 2008). Importantly, in toads, this increased dispersal is associated with a fivefold increase in the rate of range expansion. Our previous studies demonstrate accelerated range advance (Urban et al., 2008), spatial assortment by dispersal ability (Phillips et al., 2006) and morphological and behavioural shifts causing increased dispersal rates in vanguard populations (Phillips et al., 2008; Alford et al., 2009). The genetic basis of these trait shifts, however, has not been investigated. In this paper, we attempt to remedy this situation using a quantitative genetic approach.

Cane toads were introduced into northern Australia (near Cairns, Fig. 1) in 1936. Since then, they have spread, largely unassisted, across 1.3 million square kilometres of northern and eastern Australia (Urban et al., 2007). In northern Australia, the rate of toad advance has accelerated steadily, from about 10 to 55 km year−1; environmental correlates do not adequately explain this shift (Urban et al., 2008). On the invasion front, individual toads move astonishing distances (displacing up to 1.8 km per night), move predominantly in straight lines and actively use roads and other cleared areas as dispersal corridors (Brown et al., 2006; Phillips et al., 2007). Compared with toads from older populations, invasion front individuals move more often, travel farther per move and move in straighter lines (Phillips et al., 2008; Alford et al., 2009). Are these differences in behaviour evolved, or do they merely represent plasticity associated with encountering new (toad-sparse) environments? Here, we use a common garden approach to answering this question. By measuring the dispersal rates of parents collected from populations spanning the invasion history, breeding those parents and then measuring the dispersal rate of the offspring, we can minimize environmental effects on dispersal ability and estimate the heritability of dispersal. If these changes in dispersal ability are evolved, we would expect a genetic shift in dispersal between populations and we would expect dispersal to be heritable.

image

Figure 1.  Localities from which parent toads were sampled. Localities span the cane toads’ invasion history of northern Australia: Cairns (colonized 1936), Normanton (1966), Borroloola (1988) and Timber Creek (2006). All individuals were radiotracked at Middle Point, near Darwin.

Download figure to PowerPoint

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Cane toads were collected from four populations spanning the invasion history (Fig. 1): Cairns (colonized 1936), Normanton (1966), Borroloola (1988) and Timber Creek (2006). Following collection, parents were taken to a holding facility at Middle Point (near Darwin), maintained in captivity for 2 months and then released in stages (balanced by population) and located daily using radiotelemetry (see Phillips et al., 2008 for details). Each toad was measured for body size [snout–urostyle length (SUL)] and then tracked over five nights of movement before being collected and placed back into captivity.

The following year, all reproductively ready parents were mated in the laboratory. Spawning was elicited using a dose of leuprorelin acetate (Lucrin; Abbott Australasia, Kurnell, Australia – a synthetic gonadotrophin: 0.75 mL dose of 0.25 mg mL−1), and where possible, we attempted to mate one female with two males (one from her population and one from another population). We also attempted, where possible, to mate each male to at least two females (within and between populations). Multiple mating was achieved by observing the spawning process, removing the first male, washing the female thoroughly in clean water and removing any external egg strings, and then placing the other male on the female’s back and placing the pair into a new container (technique described in Howard et al., 1994). Thus, for many (14 of 20) females, multiple clutches and population crosses were obtained (Table 1). Similarly, nine of 20 males also obtained multiple clutches. For logistical reasons (primarily to maximize the number of dams), breeding was conducted at four separate times approximately a month apart. This resulted in four cohorts of offspring.

Table 1.   Sample sizes by population and mean phenotypes of radio-telemetered toads. Source populations include both parents and offspring in calculations of means, whereas cross-populations were only represented by offspring. Occasionally, sires sired multiple clutches within populations resulting in discrepancies between number of dams and sires at some sites. Figures in parentheses represent one standard deviation.
Transect distance (km)n (Dams)n (Sires)n (offspring)Mean log(daily displacement)Mean snout–urostyle length
0 (Cairns)54222.57 (1.186)87.05 (8.731)
505 (Normanton)44153.32 (1.247)86.2 (9.202)
506.5 (Cairns × Borroloola)76323.01 (1.298)76.06 (4.927)
818 (Cairns × Timber Creek)65252.67 (1.251)78.83 (5.661)
1013 (Borroloola)55213.16 (1.259)86.26 (14.013)
1070.5 (Normanton × Timber Creek)53173.43 (1.437)78.9 (4.782)
1636 (Timber Creek)22123.52 (1.699)84.57 (14.657)

For each cohort, following spawning and successful hatching of eggs, 50 tadpoles from each clutch were raised outdoors in 1000-L containers. Each container had four baskets immersed in it so that multiple clutches could be kept per container. Multiply-sired clutch pairs were split across baskets within the same large container (so that sire effects were not inflated by environmental differences). Tadpoles were fed boiled lettuce ad libitum and checked daily for metamorphosis.

Immediately following metamorphosis, the first 15 metamorph toads from each clutch were collected into clutch-specific containers and fed on field-collected termites until the young toads were 0.2–0.3 g in size (about 1 week post-metamorphosis). At this size, juvenile toads were weighed, measured, given individual toe-clips and placed into mixed-clutch rearing containers (floor area 400 × 600 mm) at a density of ten individuals per container. Clutch mixing was haphazard at this point, dependent upon the rate at which tadpoles were emerging, but each container always had a mix of clutches. Juvenile toads were raised intensively for the next 2 weeks and were weaned from termites to laboratory-reared cockroaches. At this stage, juvenile toads (up to ten from each clutch) were transferred into large outdoor rearing containers (floor area 1000 × 1000 mm) at a density of 50 animals per container and fed cockroaches daily.

Each month, all juvenile toads were censused, weighed and re-sorted so as to minimize the size discrepancies between container mates (and thus, avoid cannibalism). Clutch mixing within terrestrial rearing containers was thus haphazard and differed between months. At 30 mm in length, juvenile toads were transferred into lower density containers (same size but 20 animals per container) and raised through to adulthood on a diet of cockroaches. Water was provided ad libitum throughout via an automatic sprinkler system.

By the following year (late 2008), all toads that had reached a minimum size for radiotelemetry (> 70 mm SUL) were measured for body size and released in stages (balanced for mid-parent distance from Cairns). All offspring were then radiotracked over five nights (as had been performed with their parents).

Statistical treatment

The resultant breeding design is complex, and we are interested in assessing two questions: (i) how heritable is dispersal and (ii) does dispersal ability show a genetic shift associated with invasion history (i.e. distance through the transect across tropical Australia)? We assessed both of these questions using a single statistical model that accounts for these possible sources of variance and simultaneously estimates (i) the mean within-population additive genetic variance across the transect and (ii) the effect of transect distance on additive genetic values (breeding values). Our model was instituted in WinBUGS using code and model (an ‘animal model’) modified from that of Waldmann (2009). We used standardized log-transformed mean daily displacement as our measure of dispersal for each individual (di, which was approximately normally distributed). Thus, for parents,

  • image

where S is body size (also standardized), T is start date for radiotracking (incorporated to account for weather and time effects on field movement), B is an additive genetic value (a factor, the breeding value), and ε is the error in fit. Subscript i refers to the individual under consideration. T is a random factor (mean of zero), and B is a random factor with a mean proportional to distance from the transect midpoint (set to zero), where a negative distance represents populations close to the invasion source and a positive distance represents populations close to the invasion front, i.e.

  • image

where Ki is transect distance and Ai is the remaining additive genetic effect (random factor, mean of zero and coefficient fixed at one). Thus, Bi estimates the individual’s breeding value but partitions that value into within- and between-population effects. The between-population relationship is assumed linear (with the caveat that an actual nonlinear relationship will result in biased parameter estimates). Offspring had a similar model, but with five additional factors,

  • image

where C is the breeding cohort, X is whether or not the individual came from a population cross (which is confounded with paternity order in our design), N is the large tadpole-rearing container, R is basket within container, and D is the effect of dam (the maternal effect). X is a fixed factor, whereas C, N, R and D are random factors. In the case of the offspring, Bi is considered a random factor with mean equal to the mean of the breeding values of each parent, and a variance equal to half that around parental additive genetic variance (Lynch & Walsh, 1998; Waldmann, 2009):

  • image

where BD and BS are the breeding values of an individual’s dam and sire, respectively (estimated in first equation). This error structure should exclude all but heritable variance from our estimate of B.

We fitted this model using minimally informative priors in WinBUGS (Lunn et al., 2000; Waldmann, 2009). Convergence was assessed using three chains with randomly generated initial values. Convergence was reliably achieved within 50 000 iterations, after which we took an additional 200 000 samples from each chain to estimate the posterior distributions of our parameters.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

In total, 184 individual toads were radiotracked over the two seasons. Forty of these were the parents (20 dams and 20 sires), and the remaining 144 were the offspring of those parents. These sample sizes are necessarily small relative to most quantitative genetics studies, so our errors around parameter estimates are relatively large. Nonetheless, some patterns emerge from the data. First, the effects of rearing container, basket within rearing container, radiotracking start date and cross on dispersal were negligible, because parameter estimates for these factors were small and/or symmetrical around zero (Table 2). By contrast, the effect of body size was larger and clearly positive: larger-bodied toads tended to move farther (Table 2). Breeding cohort accounted for a surprisingly high level of variance in daily dispersal of the adult toads, suggesting that differences in birth date (and thus rearing conditions) can have relatively large influence on dispersal in this species.

Table 2.   Parameter estimates for the mixed model (‘animal model’) analysis of standardized log mean daily displacement of field-tracked cane toads sampled from populations across northern Australia.
Model coefficientsPriorPosterior
MeanSDPercentile
2.5550 (median)9597.5
  1. Letters in parentheses refer to model terms (see Methods): lower case letters refer to the coefficients of the model terms with the corresponding uppercase (fixed effects), and subscript ‘var’ refers to variance associated with random factors. Parameter estimates represent the posterior distributions of a Bayesian analysis: thus, ‘Mean’ refers to the mean of the posterior and ‘SD’ refers to the standard deviation of the posterior around that mean. Posteriors were sampled using three markov chains of 200 000 iterations following a burn-in of 50 000 iterations.

Fixed effects
 Paternity order/Cross (x)N(0,1000)0.000.20−0.39−0.330.000.320.39
 Body size (s)N(0,1000)0.200.090.030.060.200.350.38
 Distance through transect (k)N(0,1000)2.41 × 10−42.23 × 10−4−2.05 × 10−4−1.28 × 10−42.44 × 10−46.01 × 10−46.72 × 10−4
Random effects (design)
 Rearing cohort (Cvar)U(0,4)0.550.740.010.020.262.272.92
 Tadpole rearing container (Nvar)U(0,4)0.130.190.000.000.070.450.62
 Basket within tadpole container (Rvar)U(0,4)0.080.080.000.000.050.230.28
 Telemetry start date (Tvar)U(0,4)0.060.060.000.000.050.180.23
Random effects (of interest)
 Maternal variance (Dvar)U(0,4)0.110.120.000.010.070.330.43
 Additive genetic variance (Avar)U(0,4)0.290.240.020.030.230.770.93
 Error variance (εvar)U(0,4)0.690.170.320.400.710.940.99
Derived estimates
 Heritability0.280.190.020.030.240.650.74

Maternal effects were also small, although not negligible, being around one-third of the size of our estimate of additive genetic variance. Relative to other random effects, then, our estimate of mean within-population additive genetic variance was large and when combined with residual (unexplained) phenotypic variance suggested a narrow sense heritability of approximately 0.24 for log mean daily displacement in this species (albeit with a large 95% credible interval: 0.02–0.74, Table 2, Fig. 2). This heritability estimate is the mean across all four populations, but with variance due to clinal shift in breeding values between populations removed. Thus, on average, there does appear to be heritable variance within toad populations for dispersal. Selection on dispersal can, thus, produce evolved change.

image

Figure 2.  Prior and posterior distributions for estimated of mean heritability across the transect. Peaked prior resulted from flat priors for additive genetic and error variances. Figure generated from 200 000 samples of each distribution.

Download figure to PowerPoint

Has such change occurred during the invasion of toads across tropical Australia? Our estimate of the trend in breeding value between populations (k) suggests that such change is likely to have occurred (Table 2, Fig. 3). The slope estimate for between-population change was positive (indicating a genetic shift towards increased dispersal with distance from the introduction point), but the 95% credible interval does overlap zero. Inspection of the posterior distribution suggests that 86% of the posterior distribution sits above zero (Fig. 3), which is our level of support for the hypothesis of a genetically driven increase in dispersal rate during the toads’ invasion of tropical Australia.

image

Figure 3.  Prior and posterior distributions for the effect of distance from Cairns (i.e. distance from the site of the cane toads’ initial release in Australia) on breeding values for dispersal in cane toads across northern Australia. Dispersal was measured as the standardized log mean daily displacement of radiotracked toads. Eighty-six per cent of the distribution falls above zero, indicating the level of support for the hypothesis that the breeding values for toads’ dispersal rates have increased during invasion history. Figure generated from 200 000 samples of the posterior distribution.

Download figure to PowerPoint

Based on the mean of the slope posterior (k, Table 2), the breeding value of log(mean daily displacement) in toads increases 0.39 (95% confidence interval: −0.34–1.12) standard deviations between Cairns and Timber Creek. Our previous work comparing dispersal between frontal and older populations of field-collected toads (Phillips et al., 2008) revealed a shift of 1.02 standard deviations in log(mean daily displacement) between these same populations. Thus, our observed shift in phenotype in a large sample of field-collected animals is higher than, although within the confidence limits of, our estimate for the genetic shift in this trait.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Our results suggest, albeit tentatively, that cane toads on the expanding front of their invasion through tropical Australia have evolved an increase in dispersal ability. Dispersal rates of toads raised in a common-garden environment differed depending on the population from which their parents had been collected (Table 1, Fig. 4). On average, toads inheriting the genetic composition of invasion-front populations dispersed 0.39 standard deviations farther per day than did toads from long-established populations. Additionally, mixed model variance partitioning indicates that log mean daily displacement in the field is a heritable trait. In combination with the demonstration of ongoing spatial assortment by dispersal ability on the invasion front (Phillips et al., 2006), these results suggest that spatial selection and a concomitant evolved increase in dispersal have likely contributed to the accelerated invasion rate of this species.

image

Figure 4.  Mean log(mean daily displacement) for all toads tracked during the course of this study against their distance along the transect (parents) or mid-parent distance (offspring). Distance is measured in kilometres from Cairns and bars represent a single standard error.

Download figure to PowerPoint

Our inference in regard to the genetic shift in dispersal is not unequivocal, however. The posterior distribution for the effect of transect distance suggests a reasonable (14%) chance that there has been no shift or a shift in a direction opposite to that predicted. Previously, however, we have observed a phenotypic shift in a large sample of wild-caught toads of 1.02 standard deviations across this same transect (Phillips et al., 2008). If this shift was entirely caused by genetic change, then the shift in our breeding values across the transect should be similar in size. Our best estimate of the shift in breeding values is, however, 0.39 standard deviations, although the confidence limits around this estimate (−0.34–1.12) do include 1.02. This discrepancy could, thus, be attributed purely to our small sample size. Alternatively, logistical constraints imposed other design issues that we are unable to account for. For example, the animals tracked represented a nonrandom subset (i.e. the fastest growers) of all offspring produced. Thus, if there is a correlation between growth rate and dispersal (as expected by theory; Phillips, 2009; Burton et al., 2010), our sample will be biased towards the best dispersers from all clutches and may thus be biased towards showing a smaller shift across the transect. Similarly, the parents we bred from represent a subsample of all the parents available, and so correlations between breeding propensity and dispersal, for example, may have affected our results also.

The other plausible alternative to the discrepancy between phenotypic and genetic shift across the transect is that strong environmental effects on dispersal rates (e.g. for any particular genotype, toads grown in conditions associated with the invasion front may disperse farther than toads grown in source conditions) amplify the genetic shift in wild populations. Additionally, genotype by environment interactions – such as could be caused by the evolution of density-dependent dispersal during range advance (Travis et al., 2009) – also may explain this discrepancy. Obviously, further work will be needed to discriminate between these possibilities.

Despite these caveats, our study represents one of the first empirical tests of evolution on accelerating range fronts and is the first (to our knowledge) to quantify the heritable basis of changes in field-measured dispersal rates as a consequence of range shift. The increase in field dispersal rates (of both parents and offspring) with distance from the location of the toads’ initial release in Australia and the heritable nature of dispersal in toads clearly implicate the operation of spatial selection in this system.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

We thank M. Greenlees, J. Smith and T. Calnan for assistance during the initial collection of toads, as well as with radiotracking the parents. We thank J. L. McKay for his professionalism and care with radiotracking the offspring. M. Franklin, N. Somaweera and D. Nelson provided critical ongoing assistance with the animal husbandry over the 3 years of this study. The staff at Beatrice Hill Farm (particularly E. Cox and J. Stevens) provided encouragement and logistical assistance during radiotracking. The Northern Territory Land Corporation assisted with accommodation and facilities. A. Hoffmann, J. Travis, F. Janzen and an anonymous referee provided constructive comments on previous drafts. Funding was provided by grants from the Australian Research Council (to BLP and RS).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References