The early toad gets the worm: cane toads at an invasion front benefit from higher prey availability

Authors


Correspondence author. E-mail: rick.shine@sydney.edu.au

Summary

  1. In biological invasions, rates of range expansion tend to accelerate through time. What kind of benefits to more rapidly dispersing organisms might impose natural selection for faster rates of dispersal, and hence the evolution of range-edge acceleration? We can answer that question by comparing fitness-relevant ecological traits of individuals at the invasion front compared with conspecifics in the same area a few years post-invasion.
  2. In tropical Australia, the rate of invasion by cane toads (Rhinella marina) has increased substantially over recent decades, due to shifts in heritable traits. Our data on field-collected cane toads at a recently invaded site in the Australian wet–dry tropics span a 5-year period beginning with toad arrival.
  3. Compared with conspecifics that we monitored in the same sites post-invasion, toads in the invasion vanguard exhibited higher feeding rates, larger energy stores, better body condition and faster growth.
  4. Three processes may have contributed to this pattern: (i) higher prey availability at the front (perhaps due to reduced competition from conspecifics); (ii) the lack of viability-reducing parasites and pathogens in invasion-front toads; and (iii) distinctive (active, fast-growing) phenotypes of the invasion-front toads.
  5. Nutritional benefits to individuals in the invasion vanguard (whether because of higher prey availability, or lower pathogen levels) thus may have conferred a selective advantage to accelerated dispersal in this system.

Introduction

Many natural systems are in spatial disequilibrium, with the distributional boundaries of species changing in response to biotic and abiotic factors (e.g. Phillips, Brown & Shine 2010a). Understanding the processes that operate at dynamic range edges is important for conservation biologists and managers, but the challenge is formidable. Although the additional complexities introduced by spatial disequilibrium often have been ignored in mainstream ecological-evolutionary research, mathematical models of the invasion process reveal many phenomena that might be important to ecologists and conservation biologists. For example, mutations can ‘surf’ the expanding range edge (Travis, Münkemüller & Burton 2010), and rates of range expansion typically increase through time (Travis & Dytham 2002; Phillips et al. 2006). Even in the absence of natural selection, alleles that confer higher rates of dispersal tend to accumulate at expanding range edges, accelerating the invasion (Shine, Brown & Phillips 2011). A growing list of examples document dispersal-enhancing traits in frontal populations of invasive species, ranging from Canadian pine trees (Cwynar & Macdonald 1987) to European butterflies (Hill, Thomas & Blakeley 1999).

Why does dispersal rate evolve upwards during biological invasions? Under the hypothesis of natural selection, traits increase in frequency because they enhance the survival and/or reproductive success of their bearers (Darwin 1859). This hypothesis would thus suggest that individuals at the expanding range edge derive some benefit – perhaps relating to access to abundant resources, with little competition from conspecifics. Exactly what form might such benefits take? Especially for long-lived animals, quantifying actual lifetime reproductive success (LRS) of individuals that differ in dispersal rates poses a major logistical challenge (e.g. Clutton-Brock 1988). Those difficulties are especially intense at a fast-expanding invasion front, because such a study would need to encompass an area so large that it would introduce regional variation in the abiotic and biotic factors that influence LRS (thereby confounding such effects with dispersal rates). We can, however, measure traits likely to correlate with LRS (rather than LRS itself). For example, we would expect positive correlations between LRS and feeding rate, energy storage, body condition and growth rate (Roff 2002). Similarly, we can circumvent the logistical problems of assessing individual variation in dispersal rate, and regional variation in fitness-influencing factors, by confining our study to a single area and looking at changes through time rather than through space. That is, we can look at how the ecological traits of organisms change, at one location, beginning when the invasion-front arrives and continuing over time post-colonization. For example, higher food availability for individuals in the invasion vanguard should be reflected in higher feeding rates of those individuals, on average, than is the case a few years post-colonization at the same site.

In this study, we measured the energy-related consequences of being in the invasion vanguard for cane toads (Rhinella marina Linnaeus 1758) in tropical Australia. This intensively studied invasion offers a robust opportunity to look at the nature of any benefits to individuals that are in the vanguard of the invasion. Importantly, the rate of toad invasion through tropical Australia has accelerated dramatically over the 75 years since the toad's introduction (from about 10–15 to 55–60 km per annum: Phillips et al. 2006), due to shifts in toad behaviour, physiology and morphology. Compared with conspecifics from long-colonized areas, invasion-front toads have longer legs (Phillips et al. 2006), move further and more often (Alford et al. 2009) and have higher endurance (Llewelyn et al. 2010). Field measurements of dispersal under standardized conditions confirm that the accelerated dispersal is due to attributes of the toads, not to local landscapes or weather conditions, and testing of the progeny of these animals (raised in captivity) has confirmed a genetic basis to the geographic divergence in dispersal rate (Phillips, Brown & Shine 2010b). Clearly, accelerated dispersal rate has evolved at the leading edge of the cane toad invasion. In this study, we attempt to identify ecological advantages for faster dispersers that might be responsible for that rapid evolutionary change.

Materials and methods

Study Site and Sampling

Our study was carried out in Australia's Northern Territory, near the city of Darwin. The area experiences a wet–dry tropical climate, with high temperatures year-round (mean monthly maximum temperature > 30 °C in every month) and highly seasonal rainfall (75% of the annual precipitation at our study site falls within the ‘wet-season’ from January to March: Shine & Brown 2008). The results presented here combine demographic data from a mark–recapture study with data on organ weights from systematic dissections of toads.

Data on Toad Encounter Rates and Demography

In February 2005, when toads first reached the study area, we initiated a mark–recapture program at Beatrice Hill Farm (12°37′S, 131°18′ E), a 2400 ha pastoral property containing a variety of habitats (savanna woodland, pasture and floodplain). Toads were collected by hand at night, measured for body length ( = snout-urostyle length, SUL) and mass, sexed (based on skin colour and rugosity, presence of nuptial pads and/or release calls), toe-clipped for individual recognition and then released at their point of capture the following evening. To quantify an index of toad abundance, we slowly drove along a 6·4-km section of sealed roadway (which forms the western boundary of the study site) between 19·00 and 21·00 h and counted the number of live toads encountered.

Data on Climatic Conditions

Any comparison among years since toad arrival is potentially confounded by concurrent changes in weather conditions (i.e. changes in toad biology are driven by changing weather, not by time since invasion). To test whether climatic conditions changed over time coincident with time since toad invasion, we obtained data from an automated weather station located 1 km from Beatrice Hill Farm. We regressed mean monthly values of minimum and maximum temperature and accumulated rainfall against time since toad arrival.

Data on Organ Masses

Samples of toads were collected for dissection at approximately monthly intervals from five collection sites (including Beatrice Hill Farm – see above) along a linear transect, ranging from the city of Darwin through to the floodplain of the Adelaide River, 70 km to the southeast. Collections began in September 2008 when toads had been present at the five sites from 34–58 months and continued through until February 2011.

We measured SUL and mass of each toad and then euthanized it by injection of pentobarbital sodium. Toads were dissected and the liver and fatbodies removed, patted dry and weighed to the nearest 0·001 g. The proportion of the stomach containing food was estimated after pushing contents firmly to one end and the stomach was then removed, emptied of food and weighed. For logistical reasons, not all organ measurements were taken from each toad. A total of 1209 toads were dissected, fat bodies were weighed from 1036, livers from 538 and the stomach occupancy estimated from 409.

Statistical Analyses

Because our data comprise time-series data on various measures, autocorrelation may be problematic for conventional least squares analyses. Therefore, our general approach in all analyses (except for changes in growth rates) was to calculate a mean value of the variables of interest for each month and then regress these means on ‘number of months since toad arrival’. This averaging decreases levels of autocorrelation (and thus, makes the tests more conservative), but to fully eliminate autocorrelation problems, we used a linear regression model that incorporated autoregressive error terms and used maximum likelihood estimation (Proc Autoreg; sas 9.1, SAS Institute, Cary, NC, USA). We lack data for a few months, so sample sizes (number of months) vary among measures. Because our data for organ masses were obtained from five sites that differed in the dates they were initially colonized by toads, analyses of these variables do not confound weather conditions with time since toad invasion (because year 2 post-invasion, for example, was a different year at one site than another).

To quantify body condition, we used residual scores from the linear regression of ln-transformed body mass against ln-transformed SUL; toads that were heavier-than-usual (relative to length) would have positive scores, and lighter-than-usual toads would have negative scores. We used the same approach to derive scores of relative liver mass, fat body mass and stomach mass (all ln organ masses regressed against ln SUL). To estimate monthly growth rates over the 61-month study period, we restricted our analyses to growth increments measured over 20–40 days (i.e. c. 1 month). In the 26 cases where an individual's growth had been measured on more than one recapture occasion, we only used a single growth period (the one closest to 30 days in length). This provided a total of 267 data points for growth rate in both SUL and mass. Because growth rates shift ontogenetically (i.e., smaller individuals exhibit faster absolute rates of growth, with growth rate slowing after maturity), we need to standardize these growth data for body size at the beginning of the growth interval. We obtained relative growth rates (i.e., growth rate corrected for initial body size) by calculating residual scores from the simple linear regression of each growth variable (SUL or mass) on body size (SUL or mass) at the time of first capture (i.e. records of SUL growth increments were regressed on initial SUL, and mass growth increments were regressed on initial mass).

Results

Toad Encounter Rates

Although we found the first cane toad at our study site in February 2005, we did not begin to encounter toads on the roads bordering the study site for another 9 months (Fig. 1). After toads first appeared in our road surveys, their numbers increased over time (Fig. 1, Table 1). Seasonal fluctuations in toad encounter rates (toads were more common during the wet-season; Fig. 1) likely reflect changes in activity levels between wet and dry seasons, rather than fluctuations in population size.

Table 1. Temporal changes in demographic and energetic traits of invasive cane toads, and in climatic variables, as a function of elapsed time since toad colonization. Analyses are linear regressions with maximum likelihood estimates. ‘Autocorrelation terms’ indicate the number of terms added to each model to eliminate autocorrelation. Boldface font shows statistically significant effects
VariableN totalN monthly meansTime effect (no. months since arrival) t, PAutocorrelation terms
Demography
Encounter rate1350614·18, 0·00012
Mean SUL500159−1·76, 0·08381
Maximum SUL500159−3·56, 0·00080
Proportion male500159−2·67, 0·00994
Proportion juvenile5001592·73, 0·00852
Energetics
Mean body condition500159−6·14, < 0·00012
Proportion of stomach occupied 40922−1·94, 0·06680
Residual liver mass 53824−0·94, 0·361
Residual fat mass103638−4·50, 0·00012
Climate
Minimum temperature230861−0·61, 0·554
Maximum temperature2314611·26, 0·214
Rainfall2274610·41, 0·692
Figure 1.

Increasing rates of encounter with cane toads (Rhinella marina) at our study site in tropical Australia, as a function of time (number of months) since toads first arrived in the area.

Climatic Data

Monthly values of mean, minimum and maximum temperature and rainfall, did not vary significantly over time since toad arrival (all < 1·26, all > 0·21; Fig. 2).

Figure 2.

Climatic conditions over the period of our study, as a function of time (number of months) since cane toads first arrived in the area. These data were derived from an automated weather station, and comprise mean monthly values of (a) minimum and (b) maximum temperature and (c) rainfall.

Population Sex Ratio and Age Structure

During the first month of the invasion, we captured 15 toads: 12 females and three males. This initial female bias quickly reversed, as most subsequent arrivals over the first year were males (Fig. 3a). This male bias then declined over time (Fig. 3a, Table 1). There was little successful recruitment during the first year of the study. A single juvenile (SUL = 50 mm) was captured 4 months after the first toads arrived, but juveniles were not regularly encountered until 12 months after the initial invasion (Fig. 3b). Overall, the proportion of juveniles contained in monthly capture samples increased over time (Table 1).

Figure 3.

Changes in the phenotypic traits of cane toads (Rhinella marina) at our study site in tropical Australia, as a function of time (number of months) since toads first arrived in the area. The panels show shifts in (a) sex ratio; (b) mean body sizes; (c) age structure and (d) maximum body sizes.

Toad Body Size, Body Condition, Feeding Rate and Energy Stores

The first toads to arrive were large adults. Maximum size then decreased through time, with a non-significant trend for a decrease in mean body size also (Fig. 3c,d, Table 1).

Mean monthly values of residual body mass (mass relative to length) were positively correlated with corresponding values of residual liver mass (= 0·84, < 0·0001) and residual fat body mass (= 0·80, < 0·0001), suggesting that the residual body mass offers a useful proxy measure of levels of energy stores in cane toads. However, the correlation between residual body mass and proportion of stomach occupied was marginally non-significant (= 0·38, = 0·08). Although all these measures decreased over time since the initial arrival of toads, only body condition and relative fat body mass decreased significantly (Fig. 4, Table 1).

Figure 4.

Changes in the phenotypic traits of cane toads (Rhinella marina) at our study site in tropical Australia, as a function of time (number of months) since toads first arrived in the area. The panels show shifts in (a) body condition, (b) relative fatbody mass, (c) percentage stomach fullness, and (d) relative liver mass. (a), (b) and (d) are calculated as residual scores from the linear regressions of ln-transformed mass values on ln snout-urostyle length.

To determine whether any of the temporal trends in the size of toads or their organs were influenced by the change in sex ratio over time, we repeated these analyses after adding sex ratio (proportion males) as a covariate. With one exception, the sex ratio covariate was not significant in any of the regression models, and the significance of the ‘time since invasion’ effect was unaltered. In the case of mean SUL, the sex ratio effect was significant, but the temporal effect remained non-significant (t1,55 = 1·81, P = 0·08).

Toad Growth Rate

Because there was no appreciable autocorrelation among relative growth rate data (all Durbin Watson P-values > 0·39), we used simple linear regressions to examine changes in growth rate over time since toad arrival. Relative rates of growth in both body length and body mass decreased significantly with time since invasion (F1,44 = 6·61, = 0·014 and F1,44 = 9·55, = 0·004 respectively; Fig. 5).

Figure 5.

Growth rates of cane toads (Rhinella marina) at our study site in tropical Australia, as a function of time (number of months) since toads first arrived in the area. The upper graph (a) shows growth in body length, and the lower graph (b) shows growth in body mass. Both variables are calculated as residual scores from the linear regression of daily growth increment to initial body size.

Discussion

At our study site in the Australian wet–dry tropics, cane toads in the invasion vanguard had higher feeding rates, larger fat bodies, better body condition and faster growth than did conspecifics at the same site in following years. Our study was based at a single site in the very extensive (c. 3000 km long) invasion front of cane toads moving across the Australian continent; so that the studies of toads expanding into other areas (especially, cooler or drier regions) might yield different results. Below, we discuss proximate mechanisms that might be responsible for the patterns that we observed. At least four mechanisms might be responsible, and could act in combination:

  1. Climatic variation. – Toad arrival may have coincided with climatic conditions more favourable for promoting toad feeding and growth than was the case in later years. For example, warmer wetter conditions in the early years of the invasion might have resulted in more optimal physiological conditions for toads, allowing rapid growth or prolonged foraging activity.
  2. Reduced competition for food. – The first toads to arrive may have had access to prey resources that were then depleted through time; or the increasing density of toads through time may have reduced feeding rates because of higher intraspecific competition.
  3. Escape from pathogens. – Parasites and pathogens lag about 2 years behind the toad invasion front, and the pathogens closest to the invasion front have the lowest effects on host dispersal rates (Phillips et al. 2010c). Any effects of parasites on toad feeding rates or energy stores thus would be minimal at the invasion front, but increase after colonization.
  4. Attributes of toads. – Toads at the invasion vanguard are the ones that have moved further and faster, and hence may be more active, vigorous animals than later arriving conspecifics (either due to prior natural selection or to ‘spatial sorting’: Shine, Brown & Phillips 2011). Even under identical conditions of prey availability, these more athletic individuals may exhibit higher rates of feeding, and thus of energy storage and growth.

The first of these hypotheses – that hydric and thermal conditions were more favourable at around the time of toad arrival than in later years – is unlikely, based on analyses of weather station data. Our analyses show no consistent shift in these parameters with time (number of months) since toad arrival (Fig. 2). In addition, the five collection sites from which we obtained data on organ mass data were colonized by toads in different years, de-confounding broad-scale weather effects from the time of toad arrival. Hence, shifting abiotic conditions cannot explain the higher feeding and growth rates of the invasion-front toads. Other ecological factors such as food availability may have shifted across years, for reasons unrelated to weather conditions or toad abundance. Lacking data on prey availability, we cannot evaluate this hypothesis.

All three of the other hypotheses (2–4 above) plausibly have contributed to the patterns that we documented. Intuition suggests that the effects of lowered competition (hypothesis 2) would be important, given rapid increases in our rates of encounter with toads over the first few years (Fig. 1) and the propensity of toads to aggregate at high densities around feeding and hydration sites (Alford et al. 1995; Lever 2001). However, experimental studies have documented only weak competitive effects on toad feeding rates (Greenlees et al. 2007; E. Gonzalez-Bernal, pers. comm.). Those studies mimicked conditions experienced by toads around buildings, with insects attracted to artificial lights. Under those circumstances, prey availability often may be too high to detect competitive effects (Greenlees et al. 2007). Away from lights, predation by toads may be more strongly affected by densities of competing conspecifics.

Escape from native-range pathogens can influence invasive populations (e.g. Mitchell & Power 2003; Torchin et al. 2003; Perkins et al. 2008; Dunn 2009). Lungworms (Rhabdias pseudosphaerocephala) did not infect cane toads at our study site until about 2 years after toads first arrived (Phillips et al. 2010c). These lungworms affect survival and locomotor performance of metamorph toads (Kelehear, Webb & Shine 2009) and reduce growth rates of adult toads (Kelehear, Brown & Shine 2011). Thus, the delayed arrival of lungworms plausibly contributed to the decrease in growth rates of toads post-colonization. Experimental infection with this lungworm reduced prey intake in metamorph cane toads (Kelehear, Webb & Shine 2009), but not in adults (Kelehear, Brown & Shine 2011). Thus, decreased feeding rates of adult toads (as indicated by lower stomach volumes) may not be due to lungworms, although reduced activity levels of toads might cause such effects. In keeping with possible effects of parasitism on host activity levels, a standardized immune challenge greatly reduced activity levels and feeding rates of captive cane toads (Llewellyn et al. 2011).

Lastly (hypothesis 4 above), the phenotypes of invasion-front toads may differ from those of later arriving conspecifics either because of prior selection for distinctive traits at the invasion front, or simply by having been filtered spatially by differential dispersal (Shine, Brown & Phillips 2011). Even when raised under identical (common-garden) conditions, the progeny of invasion-front toads exhibit higher rates of dispersal and growth than do the progeny of toads collected in long-colonized areas (Phillips 2009; Phillips, Brown & Shine 2010b). Similarly, the progeny of invasion-front adults invested less energy into immune responses than did conspecifics from populations colonized 20–70 years ago (Llewellyn et al. 2012), potentially increasing the energy available for activity and feeding. The same kinds of genetic differences might arise during the shorter time-scale spanned by this study (5 years post-invasion).

Do the higher rates of feeding, growth and energy storage of invasion-front toads translate into increased LRS? If so, such benefits may favour the evolution of phenotypic traits that accelerate toad dispersal. This hypothesis depends upon two main assumptions. The first is that higher feeding rates of invasion-front toads are at least partly due to better conditions at the front (as posited by hypotheses 2 and 3 above) rather than phenotypic differences among toads (as posited by hypothesis 4). This assumption is likely to be met. Even if spatial sorting (Shine, Brown & Phillips 2011) modifies toad phenotypes, the benefits of lowered conspecific densities and absence of viability-reducing parasites seem likely to enhance the energy balance of invasion-front toads.

The second and the more problematic assumption is that enhanced energy balance translates into higher LRS. The validity of this commonly made assumption is difficult to evaluate. Because an individual's LRS depends upon its survival and its reproductive output, many variables might influence those two critical parameters. Cane toads have complex life cycles, and variance among individuals in LRS may depend more upon mortality during the larval stage, or intrasexual competition for mates, than upon adult feeding rates, energy stores or growth rates per se. There is no feasible way to quantify LRS in an animal like the cane toad, that produces annual clutches of tens of thousands of eggs (Lever 2001), and that can disperse > 1 km per night (Phillips et al. 2007). Indirect evidence on invasion-associated shifts in LRS-related traits does not paint any clear picture. Larval competition means that lower densities may allow toad tadpoles to develop faster, and metamorphose into larger toadlets (Alford et al. 1995). However, cannibalism of freshly laid eggs by older tadpoles may eliminate all except the first-laid clutch, resulting in only a single clutch developing within a given waterbody regardless of densities of adult toads (Crossland & Shine 2011; Crossland et al. 2011). Indeed, those later clutches may provide a nutritional advantage to older tadpoles (Crossland et al. 2011). Thus, there is no compelling reason to expect a more favourable nutritional environment for larval toads at the invasion front.

Mortality rates of terrestrial-phase toads also are difficult to measure. Rates of predation on toads likely are highest at the time of initial toad invasion, before toads reduce predator abundance (through fatal toxic ingestion) and/or induce taste-aversion learning (Shine 2010). In keeping with this prediction, faster dispersing toads at the invasion front in our study site were more likely to be killed by predators (based on radiotracking: Phillips et al. 2008). The 12-month gap between the arrival of adult toads and the appearance of juvenile toads, and low number of spawning events during the initial post-colonization years (Crossland et al. 2008), suggest low reproductive output in invasion-front toads. Thus, the higher rates of feeding, growth and energy storage in invasion-front toads may be balanced by negative effects on toad survival and/or reproductive rate.

It is difficult to evaluate the generality of our results, because few studies have followed post-invasion populations through time in a single area. Most analyses of phenotypic traits as a function of invasion history have relied upon sampling of sites with differing periods post-invasion, potentially confounding temporal shifts with locality-specific effects (Phillips, Brown & Shine 2010b). Comparisons of phenotypic traits of organisms in invasion-front vs. older populations have concentrated on traits relevant to dispersal rate. Widespread differences in LRS-related traits between populations of a species in its native habitat vs. its introduced range (e.g. Siemann & Rogers 2001; Keane & Crawley 2002; Colautti et al. 2004; Bossdorf et al. 2005; Hanski, Saastamoinen & Ovaskainen 2006; Saastamoinen 2007) may have evolved during the organism's spread through its new range, rather than as a consequence of translocation per se (Phillips et al. 2010c). Anecdotal reports suggest that population densities of invasive species (including cane toads: Lever 2001) often are high for some period post-colonization, then fall. Freeland's (1986) sampling of four cane toad populations across a 50-year invasion transect in tropical Australia (at a single point in time) found that toad densities were lowest at the longest colonized site, but remained high for at least 19 years post-colonization. Toads in long-established populations were smaller and in poorer body condition than were toads in younger populations (consistent with our results), and the slope of the relationship between body size and fatbody size varied among populations of different ages.

Post-invasion declines in invader abundance suggest that studies at replicated sites, on individuals in the invasion vanguard as well as those present a few years later, might clarify processes that affect the population viability of invasive organisms. Understanding the causes of such density fluctuations may help us to develop novel methods to reduce invader densities. Mathematical modelling predicts that invasion-front populations are subject to a suite of evolutionary forces additional to those experienced by populations that are in spatial equilibrium (Travis & Dytham 2002; Phillips, Brown & Shine 2010b; Shine, Brown & Phillips 2011). Biological invasions are now virtually ubiquitous, providing abundant opportunities for researchers to select logistically tractable model systems with which to unravel those complexities. Research on invasive species thus can contribute powerfully not only to the development of new and more effective tools for conservation and management but also can themselves act as powerful tools with which to tease apart the interacting adaptive and non-adaptive processes that generate phenotypic divergence in non-equilibrial systems (Carroll, Klassen & Dingle 1998; Carroll et al. 2005, 2007; Westley 2011).

Acknowledgements

We thank Beatrice Hill Farm and the Royal Australian Air Force for access to study sites, the Northern Territory Land Corporation for housing and the Australian Research Council for funding. The research was carried out under permits issued by the Northern Territory Parks and Wildlife Commission and with approval of the University of Sydney Animal Care and Ethics Committee (L04/4-2009/3/4999).

Ancillary