1

INTRODUCTION

EVOLUTION, CONSERVATION AND CANE

TOADS IN AUSTRALIA

2

CONTEMPORARY EVOLUTION IN CONSERVATION

There is an increasing realisation amongst biologists that evolution can occur rapidly – on time scales generally associated with “ecological time”

(Stockwell et al. 2003; Thompson 1998). It is interesting that biologists have only recently incorporated the concept of “contemporary evolution” into their worldview. We have known since before the modern synthesis that, per generation, the response to selection (R) in a trait is equal to the heritability of that trait (h2) multiplied by the strength of selection (S) operating in that generation (Lynch and Walsh 1998):

R = h 2S

This apparently simple equation tells us much about the nature of adaptation, not the least of which is that it can happen rapidly; dependent upon high heritability and strong selection acting within the range of variability for the trait in question.

We have also had, for many years, empirical evidence that under the right conditions evolution precedes rapidly from strong selection in natural populations (e.g. Kettlewell 1973). Why then, has it taken so long to appreciate the potential importance of evolution acting along timescales that are of relevance to our everyday existence? This is a question for a science historian – which I am not – but perhaps one of Gould’s (1991) creeping fox-terrier clones is to blame: Someone (perhaps Darwin himself, who as it turns out, was occasionally confused about the difference between microevolution and 3 speciation), told us that evolution takes a long time, and we have been repeating this mantra to ourselves and our students ever since.

Nevertheless, contemporary evolution has slapped us in the face until we have had no choice but to pay it some attention. Predictably in hindsight, contemporary evolution first became apparent in areas of our lives that we care deeply about; our health and our food. The green revolution and its attendant barrage of pesticides was, on reflection, a thoroughly unintended experiment in evolution. Pesticides, that were initially incredibly effective became increasingly less so as pests rapidly evolved resistance (Palumbi 2002).

Similarly, the discovery of antibiotics seemed set to save us from the scourge of disease (well, most of them anyway). Over time, our magic bullets have often come to resemble peashooters; many pathogens evolving either flak jackets or the simple ability to dodge (Ewald 1994). Two of the biggest human- killers on the planet – malaria and HIV – have achieved this position of prestige precisely because of their ability to evolve at such speed that our attacks continually hit nothing but air (Gardner et al. 2002; Palumbi 2002).

We have no choice but to acknowledge evolution when it foils our plans, but we often fail to appreciate that it can also be useful. Domestication is, of course, evolution by human selection (blessed are the pigeon fanciers, for they shall convince Darwin of the truth). And recently, medical and agricultural researchers are realising that evolutionary problems require evolutionary solutions – “evolutionarily stable strategies” are the new weapons in the wars 4 against Malaria, HIV and pesticide resistance (Ewald 1994; Palumbi 2002;

Rausher 2001).

But let us step aside from directly human-oriented concerns and apply a worldview of contemporary evolution to conservation. There is little doubt that we are living in a time of accelerated extinction (Chapin et al. 2000; May et al.

1995). Species are going extinct at higher than normal rates due to the activities of our species. Like all species, we interact with the world and, as a consequence, modify it. Unlike most species however, our modifications are large and global in their reach. In the dry language of science however, our impacts are nothing more than “environmental change”, and a geological perspective tells us that environmental change is not a new phenomenon; continents have moved, climate has changed, land bridges have allowed mass invasions, and mountains have come and gone (Morrison and Morrison 1991).

Ultimately, species either evolve through environmental change or they go extinct: Even Raup’s (1993) extinction through bad luck will almost always be the result of an environmental change. The activities of humans are causing rapid, global environmental modifications; the rapidity and size of these changes are causing an increase in the extinction rate.

Understanding how to reduce or reverse the effect of human-mediated impacts on species and communities has become the province of conservation biology. Conservation biologists know that they are fighting an uphill battle and are well aware that the resources do not come close to matching the extent of the problem. Strategies and priorities thus need to be effective and as 5 efficient as possible. It is in this desperate, outnumbered rearguard that contemporary evolution is a mostly ignored, but potentially powerful ally.

It is a property of living systems that they evolve – they are inherently dynamic and we know that evolution can happen in ecological time.

Evolutionary time is, like all time, relative; in the case of evolution, time is relative to the generation time of the organism. If an environmental change occurs along a time scale that is not too rapid, a population can mount an adaptive response. In other words, if the per generation strength of selection

(S) is not too large there is a good chance of evolution instead of extinction.

Evolved adaptation is (almost by definition) the best “strategy” for the long- term persistence of a species; from a conservation perspective, whether or not a species will adapt to a change is more important than whether it suffers short- term impact.

For conservationists, overwhelmed by the sheer number of species potentially at risk from human activities, the contemporary evolution perspective suggests that the problem may be slightly smaller than it otherwise looks. An ability to evolve around environmental change may see many species rescue themselves. The challenge for conservation biologists is to understand which species are likely to exhibit rapid evolution and which categories of environmental change are likely to encourage evolution rather than extinction.

Organisms with short generation times relative to the pace of change will have evolution on their side. Similarly, environmental changes that are 6 relatively slow and cause selection on traits that are unlikely to have been previously important for fitness will encourage evolution. Conservation biologists need to start understanding where the boundaries lie in these continuums: When does an environmental change become too rapid to expect an evolved response from an impacted species? When does the generation time of an organism become too long to preclude adaptive response to an instantaneous change of a given magnitude? These kinds of questions are not necessarily simple to answer but in answering them we can have a profound impact on the size of the conservation battle. Perhaps less species are in need of

“rescuing” than we currently believe and perhaps some environmental changes are less important than others.

This thesis represents an attempt at a small piece of this puzzle. In the pages that follow, I examine the potential for Australian snake species (which have relatively long generation times of 1-3 years) to adapt to a strong and instantaneous shift in their selective environment; the invasion of a toxic prey item.

7

SNAKES AND TOADS IN AUSTRALIA

“To others who scent a ‘nigger in the woodpile’ and suggest the possibility that the toad

will, in turn, itself become a pest, we can point to the fact that nearly 100 years have

elapsed since it was first introduced into Barbados, and there it has no black marks

against its character. Experience with it in other West Indian islands and in Hawaii,

certainly points to the fact that no serious harm is likely to eventuate through its

introduction into Queensland” R W Mungomery upon returning to Australia

from Hawaii with 101 toads in 1935.

Australia has a particularly diverse reptile fauna, among which we count around 140 native, terrestrial snake species (Cogger 2000). Thirty-three of these species are typhlopids – mostly blind, burrowing eaters of ant and termite eggs.

The remaining terrestrial species belong to the colubrid, pythonid and elapid families and eat vertebrate prey. Of these, the colubrids (11 species) are relatively recent colonisers of the continent, probably arriving from south-east

Asia during the glaciations of the early Pleistocene (2 mya, Greer 1997). The pythons (15 species) and elapids (81 species) have been present much longer, at least since the early Miocene (ca. 20 mya, Shine 1991a).

Australia’s snakes (even the newer ones!) have a long history on the

Australian continent. This venerable history contrasts garishly against the history of Bufo marinus. The species was variously referred to as the “giant toad” or “marine toad” by authors of the early part of this century (e.g. 8

Kinghorn 1938), however it is now known through most of its introduced,

English-speaking range as the “cane toad” (Lever 2001) – a fitting epithet for a species that has benefited so profoundly from the sugar-cane industry.

Originally from South America, the global sugar industry, in a welter of small, poorly thought out decisions, introduced cane toads throughout much of the

Caribbean and Pacific (Easteal 1981). It was the sugar industry that brought toads to Australia in 1935; in simple wooden crates, 101 toads were shipped from Hawaii to Gordonvale in north Queensland where, it was hoped, they were to save the sugar farmers from crop-ruining beetle pests (Mungomery

1935). The plan was that the toads were to eat their way through the cane beetle populations, leaving farmers with little beetle-filled faeces and healthy crops.

The plan wasn’t a notable success and by the 1940s toads were made entirely redundant by the development of pesticides (Low 1999). By then unfortunately, farmers (eager to ride toads to bigger yields) had distributed them along the Queensland coast and the national toad population was already well beyond control (Sabath et al. 1981). Toads have done well in Australia; they now occupy well over 1 million square kilometres of the continent and are continuing to expand their range every year (Sutherst et al. 1995).

Cane toads are a member of the Bufonidae; a family that had never before occupied Australia. Thus they were a novel prey item for Australian predators (including snakes) which had coevolved with Australia’s relatively non-toxic anuran fauna (Erspamer et al. 1984). Bufonids produce a family of 9 toxins known as bufodienolides (or bufogenins); extremely potent cardiac toxins similar to the digitalis toxins of plants but unique to toads (Chen and

Kovarikova 1967). Bufodienolides were investigated for possible use as pharmaceuticals in the 1960s only to be rejected because they produced excessive side effects and had a very narrow therapeutic window; increasing doses (among other effects) reduce the heart’s ability to synchronise muscle contractions and consequently lead to fibrillation (Chen and Kovarikova 1967).

In cane toads, this toxin is synthesised and stored primarily in the large parotoid glands located above the shoulder. There is little doubt that this well- placed chemical arsenal has had a massive impact on naïve Australian predators who die attempting to consume toads (Burnett 1997; Covacevich and

Archer 1975; Oakwood 2003; Rayward 1974 and Chapter 1, this thesis).

Natural selection comes in many guises, perhaps the most obvious of which is outright death. When a toad kills a predator, that predator has had its reproductive potential severely curbed; it has been selected against. Toads thus represent a change in the Australian environment and a new agent for

“natural” selection. Furthermore, they also represent an instantaneous environmental change; toads rapidly colonise an area so are usually either absent, or present in large numbers. As a model system within which to explore the potential for populations to adapt rapidly, toads and Australian snakes may have a lot to tell us…

10

THESIS SYNOPSIS

This introduction is a loose tour of my reasons for embarking on this line of research. It has, of course, been written after the fact, but I feel that all but the broadest reasons were somewhere in the back of my head at the outset.

Perhaps neglected as an original motivation however, was the long-running observation by myself and others that in most areas with toads, some species of snakes that should be present simply are not. I wanted to know more.

Chapter one is a necessity chapter. The sad fact is that no-one has unequivocally proven an impact by toads on any element of the Australian fauna. This is not because it hasn’t happened, but because it is almost impossible to prove. Proving a population reduction in uncommon, difficult to find organisms with populations that fluctuate wildly across seasons and years is tough. I was well aware that I could spend a decade proving an impact on one species and I was not willing to do this. Chapter one thus examines a mechanism of impact and shows that the mechanism works. The assumption, of course, is that the mechanism translates into an effect, or that a car in gear with the engine on is likely to move.

Chapter two is an interest chapter. Following the finding of ridiculous levels of resistance to toad toxin in keelbacks, I was interested to see if they really could eat toads all day long, every day and not experience some effect.

Chapter three also began as an interest chapter but quickly became important, and served to remind me that it is best to understand how selection 11 is operating before one goes out to look for its effects. The initial observation came as I was weighing dried parotoid glands and skin during initial toxin extractions. I noticed that parotoid weight became a larger proportion of total skin weight as toad size increased. I immediately concluded that larger snakes were at greater risk of poisoning from toads. It took me some time to realise that snake allometry was also important and that my initial conclusion was, in fact, wrong.

Chapters four, five and six represent the question that I was really interested in from the beginning – are snakes evolving around the impact of toads? With perfect hindsight, I know that I didn’t spend enough time examining prey preference. A quick change in prey preference is probably the best and most likely adaptation to the presence of a lethally toxic prey. This is one of those ideas that is appallingly obvious after you have worked it out but, for whatever reason, is a little opaque beforehand. In hindsight, I suspect that this is the most effective and likely adaptation for predators faced with the arrival of cane toads. Further work will have the benefit of this insight.

Chapters seven and eight share the same data and similar analyses but proved to be impossible to merge. Concatenation seemed more sensible.

Chapter seven examines the effect of time since colonisation on the aspects of toad morphology that mediate their impact on predators. The result is a pattern for which I can only guess at the mechanism. The pattern is, however, interesting and important because it indicates that toads become a less dangerous meal through time. 12

Chapter eight is a chapter about potential, and seems a fitting place to end. Spatial and temporal analyses of environmental impact and species vulnerability could allow the identification of significant times and places for adaptation to occur. The potential application for such information is extremely large and I haven’t even begun to point out its scope.

CHAPTER 1

ASSESSING THE POTENTIAL IMPACT OF

CANE TOADS (BUFO MARINUS) ON

AUSTRALIAN SNAKES.*

* Published as: Phillips, B L, Brown, G P and Shine, R, 2003. Assessing the potential impact of cane toads (Bufo marinus) on Australian snakes. Conservation Biology 17: 1738-1747. 13

14

ABSTRACT

Cane toads (Bufo marinus) are large, highly toxic anurans introduced into

Australia in 1935. Anecdotal reports suggest that the invasion of toads into an area is followed by dramatic declines in the abundance of terrestrial native frog-eating predators, but quantitative studies have been restricted to non-predator taxa or aquatic predators and have generally reported minimal impacts. Will toads substantially affect

Australian snakes? Based on geographic distributions and dietary composition, I identified 49 snake taxa as potentially at risk from toads. The impact of these feral prey also depends on the snakes’ ability to survive after ingesting toad toxins. Based on decrements in locomotor (swimming) performance after ingesting toxin, I estimate the

LD50 of toad toxins for 10 of the “at risk” snake species. Most species exhibited similar and low ability to tolerate toad toxins. Based on head-widths relative to sizes of toads, I calculate that 7 of the 10 taxa could easily ingest a fatal dose of toxin in a single meal.

The exceptions were two colubrid taxa (keelbacks, Tropidonophis mairii and slatey- greysnakes, Stegonotus cucullatus) with much higher resistance (up to 85-fold) to toad toxins and one elapid (swampsnake, Hemiaspis signata) with low resistance but a small relative head size (and thus, low maximum prey size). Overall, my analysis suggests that cane toads threaten populations of approximately 30% of Australia’s terrestrial snake species. 15

INTRODUCTION

One of the most significant threatening processes for biodiversity worldwide concerns anthropogenic shifts in geographic distributions of organisms, with natural ecosystems in many parts of the world being invaded by non-native plants and animals (Mack et al. 2000; Williamson 1996). Many such invasions are likely to cause only minor and localized ecological disruption, but some feral animals cause massive degradation and in some cases widespread extinction of the local fauna and flora (Fritts and Rodda 1998;

Ogutu-Ohwayo 1999). The processes and outcomes of ecological invasion vary considerably among systems, but potentially one of the most powerful effects involves the invasion of a toxic species into a fauna with no previous exposure to such toxins (Brodie and Brodie 1999a). In such cases native predators may be unable to tolerate the novel toxin and thus die in large numbers as they first encounter the invader.

Although there has been extensive research on the ecological impacts of invading organisms, some major questions have attracted much less attention than others. In particular, it may often be true that the most important impact of a toxic feral taxon will be on predators, yet research on changes in the abundance of predators is fraught with logistical obstacles. Because they are relatively rare and mobile, simply quantifying the abundance of many predators, let alone detecting impacts on their abundances, poses a formidable problem. In the face of such challenges researchers have tended to focus on the 16 impacts of invading species on smaller more abundant organisms – typically potential prey or competitors. There may thus be a real danger that studies will accumulate showing no (or minor) negative impacts from the invading organism, and the weight of negative evidence will encourage wildlife managers to afford less priority to potential impacts of the invasion. This is a dangerous path to follow in the absence of information on the effects of the invading taxon on predators. I believe that the spread of cane toads in

Australia reveals exactly this scenario.

The cane toad (Bufo marinus, Bufonidae) is a large (up to 230 mm body length) anuran native to South and Central America (Zug and Zug 1979).

Toads were introduced widely throughout the Pacific, primarily as an agent for biological control by the sugar industry (Lever 2001); they were introduced into

Australia in 1935 (Lever 2001). Since then, they have spread from their initial release points in eastern Queensland (Qld) to encompass more than 863,000 km2

(50% of Qld: Sabath et al. 1981; Sutherst et al. 1995). Cane toads now extend into northern New South Wales (NSW) and the Northern Territory (NT) and are predicted to further increase their range, primarily throughout coastal and near-coastal regions of tropical Australia, to encompass an area of approximately 2 million km2 (Sutherst et al. 1995 Figure 1).

These amphibians can reach astounding densities in suitable habitat (up to 2138 individuals/ha: Freeland 1986). In addition, the toad possesses a formidable chemical defence system – all life-history stages are toxic (Crossland

1998; Crossland and Alford 1998; Flier et al. 1980; Lawler and Hero 1997). The 17 active principles of the toxin (bufogenins) are extremely powerful (Chen and

Kovarikova 1967) and unique to toads (Daly et al. 1987). Toads are not native to

Australia (Lutz 1971) and therefore are both novel and toxic to Australian predators. 18

Approximate Current Distribution Predicted Distribution (current climate)

Predicted Distribution (2030 climate)

800km

Figure 1. The approximate current and predicted distribution of the cane toad in Australia. Predicted distribution is shown under current climatic and 2030 global warming scenarios (after Sutherst et al. 1995).

19

Although intuition suggests that the toad invasion may have a major ecological impact in Australia, there has been limited study of this topic. The first actual impact to be noticed was that toads were becoming significant predators of the European honey bee (Apis mellifera), causing some economic loss for apiarists (Goodacre 1947; Hewitt 1956). It was not until the early 1960s however that anecdotal reports of population declines in native species became apparent: Breeden (1963) reported observations of declines in snakes, monitors

(Varanus spp.), frilled lizards (Chlamydosaurus kingii) and quolls (a marsupial carnivore, Dasyurus spp.) following the appearance of toads. This was followed by observations of declines in snakes, monitors and birds following the arrival of toads in south-eastern Queensland and northern New South Wales (Pockley

1965; Rayward 1974). Covacevich and Archer (1975) provided further evidence for the potential impact of toads on predators by collecting numerous anecdotal reports of terrestrial predators (snakes, monitors, and marsupial carnivores) dying as a consequence of attempting to ingest toads.

Quantitative data on interactions between native species and toads have become available in the ensuing years (e.g., Catling et al. 1999; Crossland 1998;

Crossland 2000; Crossland 2001; Crossland and Alford 1998; Crossland and

Azevedo-Ramos 1999; Freeland and Kerin 1990; Lawler and Hero 1997;

Williamson 1999a). Primarily for logistical reasons, these studies have focused on interactions between toads and relatively small, abundant native organisms

(primarily fish, frogs, and aquatic invertebrates). Several of these studies have concluded that the ecological impact of toads may be less extreme than might 20 be supposed from intuition or public concern (e.g., Catling et al. 1999; Freeland and Kerin 1990; Williamson 1999a). Despite all these studies, however, there are no published quantitative analyses of the potential or realized effects of cane toads on the subset of native taxa identified anecdotally more than 30 years ago as most likely to be at risk: the terrestrial predators of frogs (The one possible exception being Burnett 1997, which focussed on Varanids and

Marsupial predators, although this study too relied on anecdotal information).

Even if competition between toads and small vertebrates is minor and their role as predators on invertebrates is modest, they might still impose a massive ecological impact if they kill a high proportion of the anurophagous predators that attempt to ingest them.

Despite the fact that snakes represent the largest vertebrate group likely to be affected, there are almost no data describing interactions between toads and native snakes. Knowledge in this area comes entirely from single observations (Covacevich and Archer 1975; Ingram and Covacevich 1990; Shine

1991c), typically of the nature of “an individual of species x survived and an individual of species y died after eating a toad.” Snakes are potentially at considerable risk from toads because many Australian snakes prey upon frogs

(Shine 1991a) and, unlike birds or mammals, have few options for prey capture.

They must use their mouths to capture and consume the toad entire and, hence, cannot avoid direct exposure to toxins in the toad’s body.

Thus, there is a critical need to evaluate the severity of the probable impact of cane toads on Australian snakes. To do so, we require information on 21 two separate topics: (1) how many Australian snake species are potentially vulnerable to toads, based on their geographic distributions and dietary habits

(i.e., how many species eat frogs and live in areas that toads will occupy); and

(2) how many Australian snake species can tolerate a quantity of toxins equivalent to ingesting a small toad?

To answer these questions, I reviewed published information on distributions and dietary habits of Australian snakes and tested the ability of 10 at risk snake taxa to tolerate toad toxins. 22

METHODS

The number of snake species potentially at risk

To identify which Australian snake species might potentially be affected by the invasion of the toad, I used ecoclimatic predictions of the likely eventual distribution of toads within Australia (Sutherst et al. 1995) and published and unpublished data on the dietary composition (Shine 1991a and refs in Greer

1997; J. Webb & G. Brown, pers. comm.) and geographic distribution (Cogger

2000) of Australian snakes.

Sutherst et al. (1995) generated two maps of the likely final distribution of cane toads in Australia, one under the present climate and one under a conservative 2030 climate change scenario. The latter method produced a slightly larger predicted distribution of the toad. I used both maps for my analysis (Fig. 1) but note that even the larger potential range might be a conservative estimate because adaptation by the toad or lack of competition from congeners may increase its range outside the ecoclimatic envelope of its native range (used by Sutherst et al. (1995) to generate predictions for the

Australian invasion). Thus, my estimates of the snake species affected may be conservative.

As well as species recorded as eating frogs, I also included some species that are likely to consume frogs but for which detailed dietary information is not available. For these species, I assessed their likelihood of consuming frogs based on dietary habits of their congeners. For each species of snake recorded 23 or likely to include frogs in their diet, I estimated the proportion of the species’ range likely to overlap with that of the toad. Multiplying this percentage by the proportion of frogs in the diet of each species yielded an index (between 0–100) of the potential impact from the toad (Table 1).

Snake’s tolerance to toad toxins

I tested 10 species of native snake for their susceptibility to toad toxin.

Because populations of snakes that are sympatric with toads might have already adapted to this novel prey type, snakes were collected from areas where toads were either absent or had been present for <15 years. Table 2 lists the species studied, with information on their body sizes and localities of collection. The study taxa included four species of snakes from the family

Colubridae, one species from the Pythonidae, and four from the Elapidae.

These species were chosen because they were all identified as “at risk” and were sufficiently common at my study sites to enable collection. Animals that were obviously ill, in poor condition, or contained large prey items were excluded from the study.

I obtained toad toxin from skins of 78 freshly killed cane toads collected from the Lismore area (northern NSW). Toads were killed by freezing. I made a single extraction of toad toxin for the entire study to remove among-toad variance in toxicity and accurately control dosing. I measured freshly killed toads for snout/urostyle length, head-width, and mass. I then removed the dorsal skin (from the back of the head to the knees) including the parotoid 24 glands. This was allowed to dry at room temperature over several days. Each skin was weighed. I then blended the dried skins with 10x v/w of 40% ethanol.

This mixture was strained and the solids discarded. The resulting liquid was allowed to evaporate to 50% of its initial volume at room temperature. I recorded the final volume and then dispensed the extract into 25 mL containers and froze it. Bufogenins are stable, partially water soluble compounds with a very high evaporation temperature (Meyer and Linde 1971). My crude extract thus contains the bufogenins although it is possible that some were lost due to saturation (see Discussion).

I tested the resistance of individual snakes to toad toxin using the decrement in swimming speed following a dose of toxin (Methodology modified from that of Brodie and Brodie 1990). Each snake was encouraged to swim around a circular pool, 3 m in diameter. The circumference of the pool was divided into quarters. I recorded snakes’ speeds (with an electronic stopwatch) as the time to traverse a single quarter. Before dosing, I subjected each snake to two swimming trials one hour apart. In each trial I recorded a snake’s time over eight quarters of the pool. The fastest speed from each trial was taken and the resulting times averaged over the two trials. This yielded an estimate of maximum swimming speed before dosing (b). I also measured the snake’s mass, snout-vent length (SVL) and head-width at this time.

The following day I gave each snake a specific dose of toxin through a feeding tube attached to a syringe or calibrated micropipette. I inserted the tube into the snake’s stomach to a depth of 30–50% of its SVL. Swimming trials 25 commenced one hour after dosing. I swam each snake twice: one hour post- dose and two hours post-dose. Maximum swimming speed was calculated as before to yield an estimate of maximum swimming speed after dosing (a). I then calculated the percent reduction in swim speed (%redn) following dosing for each snake (%redn = 100 x (1-b/a)). Experiments on neonate snakes have confirmed that reduction in swimming speed following this methodology is due to the toad toxin and not the carrier fluid (Chapter 5, this thesis).

Because I collected most of the data in the field, temperature could not be rigorously controlled across trials. I kept temperature differences between before/after trials within 2ºC by running the post-dose trial at a time when the water temperature was similar to that of the pre-dose trial. Although maximum speed may vary with temperature, the repeatability of speed assays in snakes has been shown to be consistent across temperatures (Brodie and

Russell 1999). Thus, I expect the percentage reduction measure to be unaffected by temperature differences across sets of before/after trials. All animals were adjusted to the water temperature for a minimum of 30 minutes before trials commenced by placing animals in plastic boxes floated in the pool.

I subjected each species to a range of toxin doses, with the exact range based on the effects observed. A weak or zero effect in a trial meant that the dose was doubled for the next snake, a lethal effect meant that the next dose was quartered. To minimize mortality, I initially tested snakes within each species on low doses. Each snake was tested once only. Where sample size permitted, I tested multiple snakes at each dosage level. Dosage rates were 26 calculated on a volume to mass ratio for each individual (0.002 mL/g of body mass). Different dosages were achieved by dilution of the original toxin extract.

I used six initial dilution levels (0.025x, 0.05x, 0.1x, 0.2x, 0.5x and 1x) with some species later given intermediate doses. Higher doses were achieved by successively increasing the dose per mass of undiluted extract (thus 2x = 0.004 mL/g, 4x = 0.008 mL/g, etc).

This process yielded data on reduction in speed as a function of dose for each species. In all cases there was a strong positive relationship between dose and percentage reduction in speed, within the range of doses that elicited an effect. Percent reduction scores were transformed according to the following formula modified from that of Brodie et al. (2002):

y’ = ln(2/y-1) where y is the proportional reduction in speed (%redn/100). There were three instances where the proportion reduction was <0. Because these values do not transform correctly they were entered as a proportional reduction of 0.01 for the purposes of transformation (following Brodie et al. (2002)). This transformation makes it simple to estimate the dose giving a 100% reduction in speed (the

LD50, y = 1). The reason for this is that when y = 1, y’ = 0. Thus the LD50 is the

X-intercept of the regression of y’ on dose which can be estimated as −α /β , where α is the intercept and β is the gradient of the line. Least-squares regressions of y’ on dose were conducted for each species and LD50 estimates made. 27

I performed analysis of covariance on transformed mean proportional speed reduction data with species as the factor and dose as the covariate to test for differences in resistance between species. Because of the large discrepancy in doses between some species, the data violated the assumptions of ANCOVA.

I thus performed the ANCOVA on the two natural groups of resistance (low and high) to test for differences within each group. To test whether different species required different doses to achieve a similar decrement in swimming speed across all species, I performed an ANOVA with species as the factor and dose as the dependent variable. I conducted this analysis on all individual data where the decrement in speed was >20%. Selecting the data in this way ensured that I was only comparing doses within the effect range for each species.

A snake species’ vulnerability to toads will be determined not only by the amount of toxin that it can tolerate, but also by the size of anurans that it consumes relative to its own body mass. A snake that eats only very small toads might thus be able to survive ingestion, whereas a snake that takes larger prey relative to its own body size might exceed the lethal dose. Because snakes are gape-limited predators, a snake’s head size offers an index of the maximum size of prey that it can consume (Shine 1991d). I thus calculated the head width of a toad large enough to contain a potentially lethal dose of toxin for an average-sized specimen of each snake species, and compared that prey size to the head-width of this average snake. 28

For species with sufficient data I calculated the LD50 (in terms of absolute dose) for a snake of average body size. I then converted this dose into the equivalent mass of toad skin and used unpublished data on the relationship between toad body size and skin mass to calculate the size of toad that would constitute this LD50. To compare this potentially lethal minimum toad size to the size of toad that a given snake species could physically ingest, I calculated the average mass and gape width for each snake species. I then divided the

LD50 toad size (expressed as toad head-width) by the mean snake head-width to provide an index of lethal prey size relative to the snake’s physical ability to ingest a prey item of that size. That is, the head-width of a toad of size sufficient to provide the LD50 to an average sized snake was expressed as a percentage of mean gape width for snakes of each species. Percentages of

<100% mean that the snake could easily ingest a lethal-sized toad, whereas higher values make it increasingly unlikely that the snake could ingest a toad large enough to kill it.

29

RESULTS

The number of snake species potentially at risk from toads

Analysis of distribution and dietary preference of Australian snakes suggests that 49 species are potentially at risk from the invasion of the cane toad (Table 1). Of these, 26 are likely to have their range totally encompassed by that of the toad (under predicted 2030 climate change) and three have already had their range totally encompassed by that of the toad. Nine of the “at risk” species are already recognized as being threatened either on a federal or state level ( Cogger et al. 1993). Thus, the toad invasion constitutes a potential threat to 70% of the Australian colubrid snakes (7 of 10 species), 40% of the pythons (6 of 15), and 41% of the elapids (36 of 87). These at risk taxa include 9 of the 38 terrestrial species identified as being of most concern in terms of conservation status (Cogger et al. 1993).

Snakes’ tolerance to toad toxins

For most species of snake that I tested, the percent reduction in locomotor performance was highly associated with survival after ingestion of toxin: most animals with 100% reduction in swim speeds died 1–2 hours after dosing. Animals with <100% reduction generally recovered over the course of

8–24 hours. Common blacksnakes (Pseudechis porphyriacus) were an exception to this generality, with two (of four) individuals given a 0.3x dose exhibiting swim-speed reductions of only 36% and 65%, but dying 8–24 hours later. For 30 the purposes of analysis, these individuals were scored as showing a 100% reduction in speed. 31

Table 1. Australian snake species potentially affected by the invasion of the cane toad.

Species Frogs inPercent Overlap Potential impact Conservation diet(%) Current Potential Potential index status (current climate) (2030 Climate) Boidae Antaresia childreni 33 70 100 100 33 Antaresia maculosus 6 95 100 100 6 Antaresia stimsoni 8710100.8 Morelia spilota 1 43 55 64 0.64 Morelia carinata ? 0 100 100 ? R Morelia oenpelliensis ? 20 100 100 ? R

Colubridae Boiga irregularis 6 65 91 100 6 Dendrelaphis calligastra 50 100 100 100 50 Dendrelaphis punctulatus 78 63 87 95 74.1 Enhydris polylepis 30 83 100 100 30 Stegonotus cucculatus 50 79 100 100 50 Stegonotus parvus ? ? 100 100 ? R Tropidonophis mairii 97 70 100 100 97

Elapidae Acanthopis antarcticus 6 43 51 58 3.48 R Acanthopis praelongus 27 65 100 100 27 Cacophis churchilli ? 100 100 100 ? Cacophis squamulosus 6 63 100 100 6 Demansia papuensis ? 45 100 100 ? Demansia psammophis 7 16 21 23 1.61 Demansia simplex ? 0 100 100 ? Demansia vestigiata 27 87 100 100 27 Denisonia devisii 88 45 55 60 52.8 Denisonia maculata 95 100 100 100 95 V Drysdalia coronata 53 0 38 88 46.64 Drysdalia coronoides 5 0 0 16 0.8 Echiopsis atriceps ? 0 100 100 ? V Echiopsis curta 31 0 14 32 9.92 V Elapognathus minor 66 0 0 50 33 V Hemiaspis damelii 95 60 80 100 95 Hemiaspis signata 22 75 92 100 22 Hoplocephalus bitorquatus 77 72 83 94 72.38 V Hoplocephalus stephensi 11 50 75 100 11 V Notechis ater %? 0 22 56 ? V* Notechis scutatus 92 5 5 20 18.4 Pseudechis australis 20 20 31 32 6.4 Pseudechis colletti 25 77 77 77 19.25 Psuedechis guttatus 40 64 86 100 40 Pseudechis papuanus ? ? 100 100 ? Pseudechis porphyriacus 60 32 41 53 31.8 Pseudonaja affinis 2 0 14 36 0.72 Pseudonaja guttata 41 44 47 47 19.27 Pseudonaja nuchalis 4 19 27 27 1.08 Pseudonaja textilis 9 48 52 57 5.13 Rhinoplocephalus incredibilis ? ? 100 100 ? Rhinoplocephalus nigrescens 1 61 69 77 0.77 Rhinoplocephalus pallidiceps 6 45 100 100 6 Suta ordensis ? 0 100 100 ? Suta suta 3 24 30 31 0.93 Tropidechis carinatus 41 71 86 100 41

Column 1 lists the percentage of frogs in the diet of each species. Columns 2-4 give the percentage of the species’ range encompassed by the toad currently and under the predicted distribution of toads (under present climate and a 2030 predicted climatic scenario). The index of potential impact is based on the proportion of frogs in the diet and the predicted percent overlap with toads (under 2030 climate scenario). The last column states the conservation status of individual species as listed in the action plan for Australian reptiles (Cogger et al. 1993, V = vulnerable, R = rare or insufficiently known). “?” represents an unknown quantity. “%?” represents a species for which frogs constitute a portion of the diet but for which a percentage was unobtainable. * The conservation status of this species refers to a South Australian population that is unlikely to come into contact with toads. 32

In all snake species tested, a higher dose (mL toxin/g) resulted in a greater reduction in locomotor performance (Fig. 2). The estimated LD50 was approximately 55 times higher for Tropidonophis and 22 times higher for

Stegonotus, than for the other eight taxa I tested (Table 2, Fig. 1; x-axis is ln- transformed in this figure). The LD50 for Tropidonophis was 85 times higher than the lowest LD50 estimate – that for Enhydris.

The data clearly indicate two groups of taxa – those with high resistance

(Tropidonophis and Stegonotus) and those with low resistance (all others). I performed ANCOVA separately on these two groups with species as the factor, dose as the covariate, and transformed proportional speed reduction data as the dependant variable. In both cases there was no significant interaction between species and dose factors (high, F1,13 = 0.019, p = 0.89; low, F6,14 = 2.02, p = 0.13).

After removing the interaction term, both analyses suggested significant differences between species in resistance (high, F1,14 = 24.58, p = 0.0002; low, F6,20

= 3.17, p = 0.023). In the high group Tropidonophis was significantly more resistant than Stegonotus. In the low group, this result appeared to be driven by

Hemiaspis which was slightly more resistant than other species although

Fisher’s PLSD gave only one significant pair wise comparison (Hemiaspis vs.

Pseudonaja, p = 0.02). After excluding percent reductions <20% there was no significant difference in the percent reduction scores between species (F8,91 =

1.81, p = 0.085). The dose required to achieve these similar reduction scores differed significantly among species (Fig. 3; F8,91 = 84.05, p < 0.0001). Fisher’s

PLSD confirmed that this effect was due entirely to Tropidonophis and 33

Stegonotus, which both required significantly higher doses than other species (p

< 0.02 in all cases), with Tropidonophis being significantly higher than Stegonotus

(p < 0.0001). 34

(a) Elapids + Python

100 Acanthophis

80 Hemiaspis Pseudechis

60 Pseudonaja

Antaresia 40

20

0

-20 012345678

(b) Colubrids 100 Dendrelaphis

Enhydris 80 Reduction in speed (%) Stegonotus

Tropidonophis 60 Boiga

40

20

0

-20 012345678 Ln (100 x dose)

Figure 2. Percentage reduction in speed as a consequence of toad toxin dose for 10 species of Australian snake. The x-axis, is ln(100 x dose) where dose is expressed as a concentration of toxin extract administered at a rate of 0.002mL/g. The upper graph (a) shows data for elapid and pythonid species; the lower graph (b) shows data for colubrid species. Plotted points represent the mean value for all individuals tested at each dosage level (error bars omitted for clarity). 35

0.030

0.025

0.020

0.015

0.010

Mean dose (mL/g) 0.005

0.000

s a s s s i j a i s i i i s s i i h a h r p s u h t h s c n e p d p e r o p a o a y o i l n o d d a h t e h o n u u n t m n r e e g o n e d E s s A e d a t H n i P P c e S p o A D r

T

Species

Figure 3. The average dose required to cause a reduction in speed greater than 20% for nine species of Australian snake. Mean dose is expressed as millilitres of toxin extract per gram of snake weight. Error bars represent 2 standard errors and are too small to visualize at this scale for all species except Stegonotus and Tropidonophis.

36

Table 2. The snake species examined, collection localities, morphological statistics, and results of toxin trials.

Species Location total n SVL (mm) Weight (g) Gape width (mm) LD50 Dose (mL/g) Absolute (mg) Percent of Gape Antaresia childreni Humpty Doo, NT 5 892.5 (33) 227.6 (34) 15.5 (0.6) 0.816 19.25 64.44 Boiga irregularis Casino, NSW 1 1210 (-) 299 (-) 21.9 (-) - - - Dendrelaphis punctulatus Lismore, NSW 10 984 (122) 218 (61) 19.7 (2.9) 0.744 17.51 48.83 Enhydris polylepis Humpty Doo, NT 20 611 (19) 113 (12) 13.2 (0.3) 0.448 10.56 65.17 Stegonotus cucullatus Humpty Doo, NT 18 1041 (45) 303 (37) 18.6 (0.8) 15.016 353.49 111.32 Tropidonophis mairii Humpty Doo, NT 27 550 (20) 84 (8) 12.5 (0.5) 37.992 894.35 185.54 Acanthophis praelongus Humpty Doo, NT 20 418.5 (22) 117.7 (20) 20.1 (1.1) 0.774 18.24 48.91 Hemiaspis signata Casino, NSW 3 455 (7) 32 (2) 9.4 (0.1) 0.768 18.07 107.7 Pseudechis porphyriacus Casino, NSW 28 900 (41) 347 (51) 22.0 (0.8) 0.692 16.29 42.99 Pseudonaja textilis Lismore, NSW 3 1225 (115) 592 (220) 24.9 (1.6) 0.572 13.46 36.58 Total n refers to the number of individuals tested for toxin resistance (numbers for morphological measures were different in some instances). Numbers in parentheses represent standard errors. Gape width is the distance across the head at the hinge of the jaw. The LD50 for each species is expressed as (1) the dose of toxin per body mass of snake (μL/g), (2) the absolute dose based on the weight of an average individual, expressed in milligrams of dried toad skin equivalent, and (3) as the percentage of the average snake’s head-width that a toad’s head width, whose size is sufficient to provide the absolute dose, represents (see text for details).

31 37

DISCUSSION

Cane toads have been spreading rapidly through Australia for more than

60 years, and warnings of their possible ecological impact on the native fauna have been voiced throughout that period (Lever 2001). Despite the clear inference from anecdotal reports that terrestrial predators were the component of Australian ecosystems most likely to be affected by the toads’ arrival, research on toad impacts has been dominated by studies of the effects of toads on potential prey items, competitors and aquatic predators. Several such studies have suggested that toad impacts are likely to be less severe than had been predicted, and these results have been interpreted to mean that toads may pose less of a conservation disaster than anticipated by ecological doom-sayers

(e.g. Freeland and Kerin 1990). Unfortunately, this conclusion is misleading: a lack of effect at lower trophic levels tells us nothing about potential impacts on predators, the component of the fauna most likely to be affected.

Why have previous workers focused on taxa other than terrestrial predators, despite anecdotal reports of major mortality events in native terrestrial predators (snakes, monitors, marsupial carnivores) following toad arrival? Logistical difficulties in quantifying the abundance of large vertebrate predators are the most likely reason, especially when combined with high levels of stochasticity in resources and thus, predator populations, in many

Australian habitats (Flannery 1994). Even if we cannot measure abundances accurately in the field, however, we can study the vulnerability of predators in 38 a laboratory setting to assess the likely result of encounters between a predator and a toad. The clear result from my analysis is that the invasion of cane toads is likely to have caused, and will continue to cause, massive mortality among snakes in Australia.

The methods I used to estimate vulnerability, based on geographic distributions and dietary composition, are crude and subject to several sources of error (most leading to a conservative bias). Notably, there is still uncertainty about the eventual distribution of cane toads within Australia, and we do not know how a given proportion of amphibian prey items within a particular snake species’ diet will translate into prey preferences within specific populations or among individuals within any given population. Obviously the impact of toads will be different if a 50% utilization of anuran prey is due to

50% of individual snakes eating only frogs whereas the other individuals do not attempt to consume this prey type, as opposed to the (more probable) scenario where all individuals within the population prey upon anurans as well as other prey.

Nonetheless, my analysis indicates that a high proportion of Australian snake species are potentially at risk from toads (Table 1). Although it is possible that habitat differences will reduce contact with toads for some species, the fact that the toad is an extreme generalist in Australia and can be found in most habitats (Lever 2001) suggest that this factor will be relatively unimportant. My calculations probably underestimate vulnerability for many of the taxa listed as feeding on frogs only infrequently. Low percentages of 39 anurans in the diet do not necessarily equate to low potential impact for two reasons. Firstly cane toads typically attain higher population densities than most native frogs; thus, any individual snake prepared to attack an anuran prey item is likely to encounter a toad. Secondly many snake species exhibit ontogenetic and or sex-based shifts in prey preference, such that certain size/sex classes within a population consume a higher proportion of anurans than do other size/sex classes. For example, both Pseudonaja textilis and Boiga irregularis display an ontogenetic shift from ectothermic to endothermic prey

(Savidge 1988; Shine 1989; Shine 1991c). Sexual divergence in prey composition has been recorded in A. praelongus, P. porphyriacus, and B. irregularis and is likely to be widespread across many snake species (Shine 1991b, Pearson &

Shine 2002). In all these cases, the percentage of frogs in the diet (averaged across all individuals) may lead to an underestimate of the likely impact of toads on a population.

My laboratory studies on the effects of toad toxins on snake locomotor speeds are also likely to have underestimated the severity of effects from ingesting entire toads. Firstly I only extracted toxin from the dorsal skin of toads. Toxin that is present in the ventral skin and internal organs was not included in the extraction. Secondly the extraction process is unlikely to have been 100% efficient and some toxin will have been lost. Therefore the actual lethal dose in terms of toad size is likely to be even lower than those listed in

Table 3. It is also important to note that many snakes will take multiple prey 40 items. My calculations are limited to the effect of a single prey item. Once again I am underestimating the potential impact on an individual snake.

Nevertheless, it appears that most species of snakes can easily ingest a single toad large enough to be fatal (Table 2). This result reflects the facts that:

(1) most of the snakes I tested were severely affected even by small amounts of toad toxins; (2) even small cane toads contain considerable toxin; and (3) most snakes can swallow prey items that are relatively large compared to their own body mass. In this respect, broad-headed snakes (such as Acanthophis and

Hoplocephalus) are at higher risk than relatively small-headed taxa.

Interestingly, the slightly higher resistance of Hemiaspis coupled with it’s small relative head-width suggest that this species will be less affected by toads (LD50 as % of gape width = 107%). Nevertheless, the clear result from these analyses is that most of the snake species that I tested are likely to be at substantial risk when cane toads invade their habitat.

Most of the snake species I tested exhibited low (and relatively similar) tolerance to the toxins of the cane toad (Table 2). The most striking exception in this respect was the keelback T. mairii. This species has been reported previously to ingest toads without ill effects (Covacevich and Archer 1975), but other authors have reported that keelbacks sometimes died after eating toads

(Ingram and Covacevich 1990; Shine 1991c). A survey of wild-caught keelbacks indicated that toads constituted only a small proportion of prey items inside alimentary tracts of these snakes (Shine 1991c). My data support the notion that

T. mairii is extremely resistant to toad toxin and, hence, that individuals of this 41 species are unlikely to die as a consequence of ingesting a toad. The only other snake species reported to tolerate ingestion of toads is the treesnake,

Dendrelaphis punctulatus (Covacevich and Archer 1975), but my data argue against this possibility; several individuals died after relatively small doses

(Table 2).

Although both keelbacks and slatey-greysnakes are predicted to survive the ingestion of a toad (Table 2), this does not mean these species are capable of eating toads on a regular basis. Physiological costs associated with neutralizing the toxin may entirely negate any energetic benefit associated with the consumption of the prey. Alternatively, the toxin may have a chronic cumulative effect. In the laboratory, keelbacks maintained exclusively on a diet of toads lost condition and died (Shine 1991c). It is also important to remember that my methodology removed among-toad variance in toxicity. It is entirely likely that some toads are more toxic than others and thus the outcome of an individual encounter may vary from predictions made here.

Both of the snake species that show high levels of resistance to toad toxin are colubrids. This family is believed to be a recent (post mid-Miocene but probably as late as the Pleistocene) invader of the Australian continent (Cogger and Heatwole 1981; Greer 1997). Recent ancestors of Australian colubrids are likely to have been sympatric with Bufo in south-east Asia, raising the possibility that some Australian colubrids may be pre-adapted to bufonid toxins. Tropidonophis spp. in south-east Asia prey on native bufonids (Malnate and Underwood 1988), and a genus closely related to Stegonotus in central 42

China contains at least one species that preys on toads (McDowell 1972; Pope

1935). The three other colubrids I tested (Enhydris polylepis, Boiga irregularis and

Dendrelaphis punctulatus) showed low levels of resistance, dismissing the possibility of a familial-level divergence in resistance to toad toxins.

Most of the snake species I tested showed a similar response to toad toxin on a dose per unit mass basis (Fig. 3, Table 2). This result is made more striking by the fact that snakes tested cover a broad phylogenetic span (three families). It thus seems likely that most Australian snake species will show a similar low level of resistance. However the significance of this common and low resistance level must be assessed in relation to the behaviour, ecology, and morphology of each snake species. Specific factors important in the interaction include foraging behaviour, habitat preference and the ability of snakes to learn or acquire resistance. Further research is currently underway to assess these factors – particularly the possibility of an adaptive response.

The maximum relative prey size of each species does appear to mediate the impact of toads. For example Hemiaspis signata, despite exhibiting a similar

LD50 estimate to susceptible species, is less likely to be affected because individuals can ingest only small toads relative to their body mass (Table 3).

Acanthophis praelongus on the other hand is capable of eating much larger toads than is required to provide a lethal dose, and is thus predicted to be badly affected.

My data suggest that many species of Australian snake are likely to be adversely affected by the invasion of the cane toad. The exact magnitude of the 43 effect will depend on factors specific to each species and whether or not populations can mount an effective adaptive response. However it seems prudent to treat the invasion of the cane toad as a serious threat to many populations of frog-eating snakes. Some of the species listed in Table 1 are already regarded as threatened and it would be wise for wildlife managers to give serious consideration to the impact of the cane toad on these species in particular. More generally, we should not allow logistical impediments to discourage work on the components of natural systems most likely to be affected by alien organisms.

ACKNOWLEDGEMENTS

I am extremely grateful to J. Hayter and E. Bateman for encouragement and invaluable assistance with the collection of animals. S. Hahn provided helpful advice regarding the extraction of toad toxins. G. Brown collected the

Enhydris dataset and R. Shine offered statistical advice and reviewed an earlier version of the manuscript. J. Webb kindly provided access to an unpublished manuscript. I would also like to thank the staff at Beatrice Hill research farm for their generous hospitality. Funding for this project was provided by grants from the Australian Research Council and The Royal Zoological Society of New

South Wales.

44

CHAPTER 2

SUBLETHAL COSTS ASSOCIATED WITH THE

CONSUMPTION OF TOXIC PREY BY

AUSTRALIAN KEELBACK SNAKES

45

ABSTRACT

The value to a predator of a prey item depends not only on the nutritional content of the prey but also on accessory costs such as the time, effort and risk needed to overpower and consume the prey, and any negative consequences of toxins in the prey.

Because snakes take relatively large prey, such sublethal costs associated with poor prey choice are likely to have direct fitness consequences. I examine the costs to Australian keelback snakes (Tropidonophis mairii) associated with consuming cane toads (Bufo marinus). Cane toads are an introduced species in Australia and are highly toxic; keelbacks are one of the only species of Australian snake that can consume toads without dying. Nonetheless, snakes took longer to consume toads than native frogs and toad toxin reduced snake locomotor performance for up to six hours after ingestion. These effects are likely to increase a snake’s vulnerability to predation. Although I found no significant nutritional cost associated with the consumption of toads the additional risks from eating this prey type make toads less “profitable” than native frogs. Snakes may therefore be under selection to delete such items from their diets. 46

INTRODUCTION

Research into the nutritional aspects of prey choice has a long history, spurred on by optimal foraging theory and its attendant controversies (Perry and Pianka 1997). As optimal foraging theory matured however, it became clear that factors other than nutritional benefit were also important in the evolution of optimal foraging behaviour. For example, processing time is also important (Schoener 1971) as are any other factors that reduce the predator’s benefit from consuming that type of prey (e.g. Demott and Moxter 1991;

Downes 2001). In short, a predator’s choice of prey potentially has subtler implications for fitness than are captured by the examination of calories alone.

The act of prey acquisition often carries with it many potential risks and this risk-taking is likely to have fitness consequences. Interestingly, theoreticians have incorporated subtler risks such as predation and competition (e.g. Brown

1999; Houston et al. 1993) but empirical work has lagged behind (Perry and

Pianka 1997).

The riskiness of prey acquisition is perhaps nowhere better demonstrated than by snakes, which tend to eat large prey and are relatively defenceless during the ingestion phase (Cundall and Greene 2000).

Additionally, many snakes consume toxic prey – usually amphibians – many of which contain toxic skin secretions (Daly et al. 1987; Duellman and Trueb 1994).

Snakes thus face not only the obvious risks associated with subduing large prey and being vulnerable to predation while ingesting this prey but they also often risk being poisoned. Specialised diets and venom may well be evolutionary 47 responses to these risks (Greene 1997) and hence testament to their importance to snake fitness.

Much of the literature dealing with the costs of toxic prey revolves around herbivores and the costs associated with the ingestion of phytotoxins

(e.g. Agrawal and Klein 2000; Demott and Moxter 1991; Guglielmo et al. 1996).

In these systems optimal choices are often inferred to depend almost solely upon the nutritional value of the food and the energetic cost of dealing with the toxin. Fitness consequences associated with other aspects of the organism’s environment, such as predation, are often ignored (Bernays and Graham 1988;

Dicke 2000). Few studies then, examine the potential consequences of toxic prey choice from both a nutritional and ecological perspective.

If we are to understand foraging behaviour, then optimal foraging should be defined as that which maximises lifetime fitness (Perry and Pianka

1997). As such, factors other than nutrition alone must be considered. Snakes, offer ideal study organisms in which to examine this problem – poor choice of prey not only has nutritional consequences but may also increase a snake’s chance of injury or predation. In this paper I examine implications of prey choice in the Australian keelback snake (Tropidonophis mairii) in terms of risk as well as energy.

48

METHODS

Study species

The keelback is a small, crepuscular, non-venomous colubrid that feeds almost exclusively on anurans (Cogger 2000; Shine 1991c). It is the only

Australian snake known to be regularly capable of ingesting the introduced cane toad (Bufo marinus) without dying (Phillips et al. 2003). Cane toads are highly toxic and highly successful invaders of the Australian continent, having colonised more than 1 million square kilometres since their introduction in 1935

(Lever 2001; Sutherst et al. 1995). Once toads become established they can reach astounding densities, often becoming the most common anuran species in an area (Freeland 1986). Because of this, toads are often the most common prey available to anurophagous snakes such as the keelback. In this study, I evaluate the consequences to keelbacks of consuming cane toads instead of native prey

(rocket frogs, Litoria nasuta). Rocket frogs are a common, relatively non-toxic species (Erspamer et al. 1984) that figure prominently in the natural diet of the keelback (Shine 1991c).

I tested three possible non-lethal effects associated with ingesting toads vs frogs. First, I tested for non-lethal performance effects and the duration of such effects, using locomotor performance as an indicator of performance.

Second, I tested the possibility that toads require greater handling time

(consumption time). Third, I tested the possibility that toads make “poorer” meals from an energetic perspective. 49

Effects of toxin on locomotor performance

I tested the effect of toxin on the locomotor performance of 12 neonate keelbacks hatched in the laboratory as part of an on ongoing study into the ecology of this species. Neonate keelbacks are likely to be at the highest risk from toads because small snakes tend to eat relatively larger prey items and hence consume relatively larger doses of toxin (Chapter 3). Additionally, young keelbacks suffer extremely high mortality in the field, probably as a result of predation (G Brown pers. comm.). I assigned three keelbacks to each of four treatment groups, each receiving different doses of toxin. Toxin was extracted from the skin of 78 toads. Details of the toxin extraction procedure can be found in chapter 1.

Locomotor performance was assayed using a methodology modified from that of Brodie and Brodie (1990) and used extensively in chapter 5.

Individuals were swum along a 2m swimming trough and were timed with an electronic stopwatch over three consecutive 50cm segments of the trough.

Animals were encouraged to swim by tapping them on the tail. Water temperature was maintained at 23±1oC.

A swimming trial consisted of two consecutive laps of the trough. This yielded 6 measurements of swim speed over 50cm of which only the fastest was retained. All animals (1-2 days post hatching) were initially subjected to three swim-trials one hour apart. This yielded three maximum sprint speed times 50 that were averaged to generate the pre-dose estimate of maximum swim speed

(expressed as the time taken to cover 50cm, b).

On the following day, snakes were given a specific dose of toxin or control solution. The doses were either 25μL, 50μL or 100μL of toxin or 100μL of control solution for each treatment group. These doses all fall within the range that a snake would experience from ingesting a single toad (Phillips et al.

2003). Control solution was prepared in a manner identical to the toxin extraction but without the addition of dried skin. Dosing was achieved by use of a micropipette attached to a thin rubber feeding tube. The tube was inserted into the stomach to a depth of 5cm from the snout before toxin was expelled.

Animals were observed for 1 minute following this procedure to ensure the toxin was not regurgitated. Several swim trials were then undertaken for each individual. Locomotor performance was assessed 30min, 1.5hrs, 2.5hrs, 6hrs and 18hrs post dosing. Once again, only the fastest speeds for each trial were taken. This yielded two time measurements, which were then averaged to give post-dose swim speed (again, expressed as time, a) for each time post dose. The percentage reduction in swim speed (%redn) was calculated from these times using the formula, %redn = 100 x (1-(b/a)).

Consumption time and energetic benefit

Adult keelbacks were captured by hand at Tyto Wetlands, 1 km south west of Ingham, Qld in February 2003. Snakes were housed individually in plastic tubs in a field laboratory maintained at 27oC. Each animal was given a 51 pile of straw for a retreat site and a container of water, which was available at all times throughout the experiment. A total of 21 snakes were collected. After seven days of acclimation, each snake was weighed, given an identification number and randomly allocated to one of two treatment groups. One treatment group was offered frogs (Litoria nasuta) for prey (10 snakes) and the other group was offered cane toads (Bufo marinus, 11 snakes).

Snakes were offered prey on three separate occasions separated by three days. Prey items were left in each snake’s enclosure for 24 hours. At the end of this period the cage was carefully searched for prey items and I recorded whether or not the prey item was eaten. I attempted to keep the relative amount of food per snake similar between the two groups. This was achieved by feeding each snake approximately the same relative prey mass (prey mass divided by snake mass) each feeding session. The toads I collected weighed less than the frogs on average, so often multiple toads were given to a snake to reach the equivalent mass of a single frog.

Where possible I observed snakes consuming prey and scored the prey orientation (head first or legs first) and the time it took a snake to consume a prey item (from initial grab until the snake’s mouth was completely closed with the prey item in the throat).

After the three feeding opportunities, no snake was offered food. To allow digestion of food and passage of faeces, I waited one week after the conclusion of feeding before weighing all the snakes again. During this period, all snakes were offered water as before. 52

I compared toad and frog treatments using ANCOVA with change in snake mass as the dependent variable and total prey mass as the covariate. I also compared consumption time between the treatments (with relative prey mass as a covariate) to determine whether frogs were consumed faster than toads.

53

RESULTS

Effects of toxin on locomotor performance

All doses of toxin given to neonate keelbacks elicited a reduction in locomotor performance; an effect that decayed with time (Fig. 1). Repeated measures analysis of variance with %redn as the dependent variable revealed a strong interaction between dose level and time (Wilks’ lambda; ≈F12,13.5 = 5.96, p

= 0.0013). This effect remained even after the control group was excluded from the analysis as expected if all the time series converge on a similar value (see

Fig. 1 at time 18 hrs).

Therefore, I chose to perform an ANOVA at each time interval and use

Dunnet’s post hoc test to compare %redn for each treatment against that of the control. This analysis showed significant differences between the two highest dose categories and the control for the first three post-dose trials (0-2.5 hrs). At six hours post-dose, only the high dose category exhibited a significant difference and by 18 hours post-dose none of the treatment groups showed locomotor decrements significantly different from those of the control group

(Fig. 1).

54

100100

7575

High dose (100uL) 5050 Medium dose (50uL)

Low dose (25uL) 2525 Control

00 Mean reduction in speed (%) Mean reduction in speed (%)

-25-25 0 5

0 5 1010 1515 2020

HoursHours postpost dosedose

Figure 1. Effect of toxin on locomotor performance in neonate keelback snakes at three dosage levels over 18 hours. For clarity, error bars (one standard error) are shown only for the high dose and control treatments (errors were similar across all treatments). Closed symbols represent treatment groups where the percentage reduction in locomotor performance was significantly different from that of the control (Dunnet’s post hoc test). 55

Consumption time

Two snakes from the frog prey treatment died before the end of the feeding experiment for reasons unrelated to the experiment. Thus my final sample size was eight animals in the frog treatment and eleven in the toad treatment.

One individual took > 15 minutes to consume a large toad, after which it appeared unable to move (not responding to a touch on the tail) for > 3 hours.

While this individual demonstrates the potential difficulties associated with eating toads the data point was removed from further analyses because it was an extreme outlier and may bias the analysis. ANCOVA of consumption time by prey type and prey orientation (relative prey mass as the covariate) revealed no significant interaction terms (p > 0.1 in all cases). After removal of interaction terms I found a significant association between relative prey mass and consumption time (F1,13 = 47.66, p < 0.0001) and a significant difference in consumption times between the two prey types (F1,13 = 8.48, p = 0.012) with toads taking longer to be consumed (Figure 2). Prey orientation, however, had no significant influence on consumption time (F1,13 = 0.05, p = 0.83).

Additionally, no significant difference was observed in prey orientation between the two prey types (% head first: toads, 60%; frogs 40%; χ2 = 0.805, p =

0.37).

Energetic benefit 56

At the conclusion of feeding trials, all animals were weighed and their change in mass over the course of the experiment was calculated. All animals fed frogs gained mass whereas four of the eleven individuals maintained exclusively on toads lost mass. I used ANCOVA with change in snake weight as the dependent variable. Prey type, initial snake weight, and total prey weight were used as independent variables. In this model, there were no significant interaction terms (p > 0.08 in all cases). Following the removal of interaction terms, prey type had no significant effect on weight change in the snakes (F1,15 = 0.001, p = 0.98) although both initial snake weight and total prey weight had a strongly significant effect on weight change (p < 0.0005 in both cases).

57

1250

Frogs

1000 Toads

750

500

Consumption Time (s) 250

0 0 5 10 15 20

Relative Prey Mass

Figure 2. Consumption time against relative prey size for keelback snakes consuming two prey types – toads and frogs. Relative prey mass = the mass of prey divided by the mass of predator multiplied by 100. The single extreme consumption time value (time >1000 s) for an individual consuming a toad was excluded from the statistical analysis.

58

DISCUSSION

The keelbacks in this study demonstrated significant sublethal costs associated with the consumption of toads. Despite the apparent lack of energetic cost associated with the consumption of toads, these toxic anurans took longer to consume, and the toxin in toad skin inhibited snake locomotion for up to six hours.

Keelbacks are unusual among Australian reptiles in having a very high resistance to toad toxin. For example, keelbacks are more than 70 times more resistant to toad toxin than are many other Australian snakes (Phillips et al.

2003). Thus they are one of the few taxa that are able to consume toads without dying as a result. My results also suggest that keelbacks can derive energetic benefit from the consumption of toads. Although there may be a differential in energy benefit between frogs and toads, it is not striking and was not detectable with my small sample sizes.

Although keelbacks are capable of eating toads and can derive significant energy benefit from them, they do experience costs associated with toad ingestion. First, consuming a toad reduces the keelback’s locomotor performance. The extent of this cost is dose related and will thus depend upon the relative size of the toad eaten. Some doses reduced snake mobility for at least 6 hours. The biological relevance of these doses is evidenced by the observation of one snake being unable to move for more than three hours following the ingestion of a large toad during the feeding trials. Such 59 hampered locomotor performance potentially has serious implications for predator avoidance, particularly for an active, fast moving snake such as

Tropidonophis mairii.

Also, the keelbacks in this study took longer to consume a toad than to consume frogs of equal size. Higher consumption time is likely to require greater energy expenditure and more importantly, will increase a snake’s susceptibility to predation. The higher consumption times for toads may be due to several possible factors. First, toads have drier skin than frogs so more salivary lubrication is necessary for ingestion of toads. Second, toads inflate their lungs when grasped by a snake (thereby making ingestion more difficult), whereas Litoria nasuta does not. Other species of Australian frog (e.g. Cyclorana spp.) do inflate their bodies (Williams et al. 2000) and it would thus be possible to determine whether the increased consumption time was due to this aspect alone. Third, ingestion time may be extended because keelbacks affected by toad toxin exhibit muscular weakness and lowered performance. The rapid effect of toad toxin may affect the efficiency of ingestion by snakes.

These subtler costs (whatever their mechanism) are likely to have fitness consequences. There are two reasons for this: (1) Snakes run a high risk of injury in the act of subduing prey. Because snakes are limbless, all prey acquisition and handling must be initiated with their head (Cundall and Greene

2000). Because the head contains all the sensory organs it is the most vulnerable part of the snake. Despite this vulnerability, snakes often consume relatively large prey, so the risk of injury whilst subduing prey is high. Given these 60 circumstances, rapid subdual of prey is important (probably the reason many species have evolved venom or prey specialisation). Prey that take longer to subdue are more likely to cause an injury. (2) To avoid predators, most snakes depend upon crypsis and flight (Greene 1997). Catching and subduing prey will often rob an individual of the advantage of crypsis. After prey has been subdued, the process of ingestion is relatively slow (snakes eat large meals) and while ingesting prey, snakes are unable to move effectively and are unable to defend themselves. Rapid ingestion of prey thus minimises the chance of predation. Additionally, prey that are sufficiently toxic to impair locomotor performance clearly reduce a snake’s ability to flee.

Even in the absence of a direct energetic consequence of eating toads, these subtler effects are likely to have an impact on snake fitness. These “risk consequences” thus represent an additional evolutionary hurdle to any snake population utilising toads as a food resource. While relatively obvious in snakes, such subtle effects may be very common in predator/prey interactions generally and should not be overlooked. Even in the absence of obvious energy differentials between prey types, “risk consequences” may be driving the evolution of prey choice.

ACKNOWLEDGEMENTS

I would like to thank Virginia McGrath, David Fouche, Michael Kearney,

Luke Shoo, Gavin Bedford, James Smith and Kath Nash for their assistance in 61 the field. I am also grateful to Thomas Madsen and Greg Brown for providing logistical support. Richard Shine and two anonymous referees improved earlier drafts. This research was supported by grants from the Norman

Wettenhall Foundation and the Linnean Society of NSW (to BLP) and a grant from the Australian Research Council (to RS).

62

CHAPTER 3

ALLOMETRY AND SELECTION IN A NOVEL

PREDATOR-PREY SYSTEM: AUSTRALIAN

SNAKES AND THE INVADING CANE TOAD

63

ABSTRACT

Because many organismal traits vary with body size, interactions between species can be affected by the respective body sizes of the participants. I focus on a novel predator-prey system involving an introduced, highly toxic anuran (the cane toad, Bufo marinus) and native Australian snakes. The chance of a snake dying after ingesting a toad depends on the size of the snake and the size of the toad, and ultimately reflects the effect of four allometries: (1) Physiological tolerance (the rate that physiological tolerance to toad toxin changes with snake size); (2) swallowing ability (the rate that maximal ingestible toad size (i.e. snake head size) increases with snake body size); (3) prey size (the rate that prey size taken by snakes increases with snake head size) and (4) toad toxicity (the rate that toxicity increases with toad size). I measured these allometries, and combined them to estimate the rate at which a snake’s resistance changes with toad toxicity. The parotoid glands (and thus, toxicity) of toads increased disproportionately with toad size (i.e., relative to body size, larger toads were more toxic) but simultaneously, head size relative to body size (and thus, maximal ingestible prey size relative to predator size) declined with increasing body size in snakes. Thus, these two allometries tended to cancel each other out. Physiological tolerance to toxins did not vary with snake body size. The end result was that across snake species, mean adult body size did not affect vulnerability. Within species, however, smaller predators were more vulnerable, because the intraspecific rate of decrease in relative head size of snakes was steeper than the rate of increase in toxicity of toads. Thus, toad invasion may cause disproportionate mortality of juvenile snakes, and adults of the sex with smaller mean adult body sizes. 64

INTRODUCTION

An organism's body size profoundly influences a multitude of biologically significant traits, ranging from metabolic rates to habitat use, modes of locomotion and life history characteristics (Calder 1984; Schmidt-

Nielsen 1984). These kinds of effects are seen within as well as among species, although often less dramatically in intraspecific comparisons because of the smaller size range combined with developmental constraints (Schmidt-Nielsen

1984). Nonetheless, ontogenetic increases in body size will influence the way an organism interacts with its environment, modifying aspects such as reproductive capacity, competitive ability, optimal choice of food and shelter and the risk of predation (Gould 1966). In consequence, evolutionarily significant attributes of the life history (e.g., rates of mortality and reproductive success) often are closely linked to body size (Calder 1984; Stearns 1992).

Commonly, the important issue is not absolute body size, but an organism's size relative to that of other organisms. Predator-prey interactions offer some of the clearest examples of situations where fitness consequences depend upon the respective body sizes of the two participants. For example, a predator's ability to locate, capture, kill and ingest a prey item will obviously depend not only upon its own body size, but also on its size relative to that of the prey item.

Encounters between predators and highly toxic prey offer an opportunity to examine the multiple allometries affecting the outcome of predator/prey interactions. Information on this topic is of special value when 65 the interaction involves an invasive species that causes massive disruption to natural ecosystems. The invasion of the cane toad (Bufo marinus) in Australia provides such a case. Cane toads were introduced into Australia in 1935 in a failed attempt at biological control of agricultural pests (Lever 2001). They have spread their range rapidly since that time, and eventually will occupy almost one third of the continent (Sutherst et al. 1995). Toads are highly toxic, especially to Australian predators with no evolutionary history of exposure to their toxins (Chen and Kovarikova 1967; Lutz 1971). Many Australian native predators die if they attempt to ingest a toad (Covacevich and Archer 1975;

Phillips et al. 2003). Thus, toads represent a novel selective force for these predators and a massive management problem for conservation agencies (e.g. van Dam et al. 2002). At least 30% of Australia’s terrestrial snake species are at risk from these toxic invaders, and most snakes are likely to die if they ingest a toad (Phillips et al. 2003). Australian snakes are thus facing severe population declines in the presence of toads.

How does body size influence the interaction between snakes and toads?

Toads and snakes offer an excellent opportunity to study the effects of body size (allometry) on predator-prey interactions because: (1) toads are a strong novel selective force on snakes; (2) both snakes and toads span a large range in body sizes (as is true for ectotherms in general: Bonnet et al. 1998; Pough 1980);

(3) the interaction is relatively simple to quantify, because snakes are gape- limited predators that must swallow an entire prey item rather than tearing it apart; and (4) the timing and extent of the toad invasion in Australia is well 66 documented (Lever 2001). In this paper I describe allometric variation in toad toxicity, snake resistance to toad toxin, and snake head sizes. Using these data,

I examine the ways in which the respective body sizes of prey and predator influence snake vulnerability, and hence the selection pressures that toad invasion imposes on body sizes of snakes. 67

MATERIALS AND METHODS

Analysis framework

All my analyses concern allometry. Relationships between log- transformed morphological variables should yield simple ratio relationships depending upon the units of measurement for each variable (Calder 1984;

Schmidt-Nielsen 1984). Thus, null slopes are obtained by assuming that there is no change in the shape of an organism through ontogeny (i.e. isometry). Under this assumption and with both variables log-transformed, plotting mass (a cubic variable) against length (a linear variable) should yield a slope of three.

Similarly, a length variable plotted against another length variable should yield a slope of one.

Ultimately, I aim for an understanding of how snake resistance changes with toad toxicity. In calculus terms, I want to calculate:

δ(snake resistance)

δ(toad toxicity )

This allometry can be viewed as the product of a number of contributing allometries where numerators and denominators cancel out, such that:

δ(snake resistance) δ(snake resistance) δ(snake length) δ(snake head size) δ(toad size) = × × × δ(toad toxicity ) δ(snake length) δ(snake head size) δ(toad size) δ(toad toxicity )

Thus, to understand the impacts of allometry on the interaction between toads and snakes, we need information on four kinds of allometries:

68

⎛ δ(snake resistance)⎞ 1. The allometry of snake resistance against snake length ⎜ ⎟ . ⎝ δ(snake length) ⎠

Does a snake’s ability to deal with toad toxin increase or decrease in proportion to the snake’s size? The absolute quantity of toxin required to kill a snake presumably will increase with increasing body size, but the snake's vulnerability will depend upon the slope of this relationship. For example, if larger snakes can withstand higher toxin doses relative to their size, then a smaller predator may be at higher risk from ingesting a toad.

⎛ δ(snake length) ⎞ 2. The allometry of snake length against head size ⎜ ⎟ . ⎝ δ(snake head size)⎠

Do larger snakes have heads that are smaller (relative to body length) than do smaller snakes? Ontogenetic decreases in relative head size occur in many kinds of animal, but are especially important in snakes because these animals are gape-limited predators. That is, gape size determines maximum ingestible prey size (Arnold 1993; Cundall and Greene 2000). Thus, deviations from isometry between a snake's body size (which influences the absolute amount of toxin that would be needed to kill the snake) and its gape size (which influences maximum prey size, and thus the absolute amount of toxin that can be ingested) may have important implications for the toad/snake interaction.

Thus, for example, a decrease in relative head size in larger snakes might render them less likely to ingest a fatal toxin dose.

69

⎛ δ(snake head size)⎞ 3. The allometry of snake head size against toad size ⎜ ⎟ . ⎝ δ(toad size) ⎠

This allometry represents the interaction between predator and prey. If snakes of all sizes only consumed prey items that were the biggest they could physically ingest, this relationship should be approximately isometric. Usually, however, snakes eat smaller prey as well as occasional maximal-sized items

(Arnold 1993). For example, many of the prey encountered by very large snakes might be well below their maximal swallowing capacity, whereas this is less likely for smaller conspecifics. If this is the case then smaller snakes are taking relatively larger meals (and hence more toxin) relative to their own body size.

⎛ δ(toad size) ⎞ 4. The allometry of toad size against toad toxicity ⎜ ⎟ . ⎝ δ(toad toxicity )⎠

Large toads presumably contain more toxin than do small toads, but the rate at which the body size of a toad increases relative to its toxin content has strong implications for a predator that may consume small toads when young and large toads when mature.

Whether or not a snake dies after eating a toad will depend upon all of these allometries, so that we need to combine information on all of these aspects to understand the effect of snake body size on the level of risk imposed by toads. This approach can be used to investigate allometries at two levels: 70 intraspecific (are smaller or larger individuals within a given snake population more at risk?) and interspecific (are species with smaller or larger mean adult body sizes more at risk?). Below, we explain the methods by which we estimated each of the component allometries.

71

Data collection

All the snake species I examined have a substantial range overlap with toads and feed largely on anurans. Additionally, all the species I examined

(with the exception of keelbacks, Tropidonophis mairii) have similar, low resistance to toad toxin (Phillips et al. 2003) and as such, are predicted to be badly affected by the presence of toads. Keelbacks are unusual among

Australian snakes in having a very high resistance to toad toxin (Phillips et al.

2003).

⎛ δ(snake resistance)⎞ 1. The allometry of snake resistance to toad toxin ⎜ ⎟ . ⎝ δ(snake length) ⎠

To calculate this allometry we would, ideally, give the same absolute dose of toxin to snakes across a range of body lengths and measure the effect.

Unfortunately, the range of toad-toxin doses that elicit measurable but non- lethal effects on snakes is extremely narrow (Phillips et al. 2003). Thus, I opted to give each snake a size-adjusted dose and then determine if any residual correlation existed between response to the toxin and body size. Such a correlation would indicate a deviation from isometry in snake resistance.

Additionally, because I expected resistance to scale with body mass rather than length, I gave snakes a mass-adjusted dose rather than a length-adjusted dose.

The allometry of resistance against mass can be converted to resistance against length by multiplication with the allometry of mass on length, i.e.:

δ(snake resistance) δ(snake resistance) δ(snake mass) = × δ(snake length) δ(snake mass) δ(snake length) 72

Thus, for the intraspecific comparison, I first examined the allometry of resistance against mass and then measured the allometry of mass on length to arrive at an estimate of the allometry of resistance on length.

Intraspecific allometry of snake resistance to toad toxin.

Toxin was obtained from skins of 78 freshly killed cane toads collected from the Lismore area (northern NSW) and killed by freezing. A single extraction of toad toxin was made for the entire study, to remove among-toad variance in toxicity and accurately control dosing. The dorsal skin (from the back of the head to the knees) including the parotoid glands was removed and dried at room temperature over several days. The dried dorsal skin and parotoid glands were weighed and then blended with 10x v/w of 40% ethanol.

This mixture was strained and the solids discarded. The resulting liquid was allowed to evaporate to 50% of its initial volume at room temperature. The final volume was recorded and then aliquoted into 25ml containers and frozen.

I used this extract to assess toxin resistance of two species of snake: the keelback, Tropidonophis mairii (Colubridae) and the red-bellied blacksnake,

Pseudechis porphyriacus (Elapidae). These two species span the entire range of toxin resistance among Australian snakes with P. porphyriacus having the common, low resistance and T. mairii having very high resistance (Phillips et al.

2003). Tropidonophis mairii were collected from the Adelaide River floodplain 60 km east of Darwin, Northern Territory and P. porphyriacus were collected from various localities within New South Wales and Queensland. Within each 73 species, each snake was given the same dose of toad toxin extract relative to body mass. However, the dose differed between species and was calculated from earlier trials (Phillips et al. 2003) to maximise my ability to detect variation between individuals in resistance whilst minimising mortality. Thus, T. mairii were given 24μL/g, and P. porphyriacus 0.4μL/g body mass. To quantify physiological tolerance, I measured locomotor speeds of each snake before and after receiving this standard dose of toad toxin (see Chapter 1 for details of these methods). Because each snake was given the same dose of toxin relative to its body mass, we can expect that there will be no relationship between body mass and response to toxin (i.e., % locomotor decrement). To look for any such relationship (which would indicate a deviation from isometry), I regressed locomotor decrement against mass for each species.

Intraspecific allometry of snake mass relative to body length.

I obtained mass and length data for four species of snake including the two species that were tested for toxin resistance (above). The additional species were the swampsnake (Hemiaspis signata) and the green treesnake (Dendrelaphis punctulata). These data come from snakes collected in various localities in

Queensland and New South Wales (including the snakes used in the toxin resistance trials). I used reduced major axis regression (Sokal and Rohlf 1995) to calculate the slope coefficients of mass against snout-vent length (SVL) for each species. The observed relationship from the data was tested against the 74 null expectation (slope = 3) using simple t-tests, with standard errors taken from the appropriate least squares regression.

Interspecific allometry of snake resistance to toad toxin.

Data from Phillips et al. (2003) strongly suggest that most Australian snakes have similar levels of mass-adjusted resistance to toad toxin. The species tested by Phillips et al (2003) spanned a broad range of body sizes (420-

1225mm SVL) so for the purposes of the interspecific comparison I assume that resistance scales isometrically with body length.

⎛ δ(snake length) ⎞ 2. The allometry of snake length against head size ⎜ ⎟ . ⎝ δ(snake head size)⎠

I measured body and head size for preserved museum specimens of five species of snake: two colubrids (keelbacks T. mairii, green treesnakes D. punctulatus) and three elapids (swampsnakes H. signata, rough-scaledsnakes

Tropidechis carinata, red-bellied blacksnakes P. porphyriacus). All specimens were measured for snout-vent length (SVL), head-width (HW; across head, where supraoculars meet parietals) and jaw length (JL; jaw hinge to tip of snout).

Within each species, I assessed allometry in head/body dimensions by regressing log-transformed SVL against head size variables, using RMA regression (null slope = 1). 75

Because allometry of traits within a species may differ from patterns across taxa, I also examined relative head dimensions interspecifically, across 30 species of Australian snakes from four families. For this analysis I used published data on mean SVL and head length (head width data were not available, Shine 1991b); head width data were unavailable. These data were plotted and allometric coefficients calculated as for the intraspecific analyses.

⎛ δ(snake head size)⎞ 3. The allometry of snake head size against toad size ⎜ ⎟ . ⎝ δ(toad size) ⎠

For the calculations that follow, I assume that this relationship is isometric (i.e. slope = 1). While it is reasonable to assume that maximum ingestible prey size scales isometrically with head size, actual prey size taken by snakes is likely to deviate from isometry. This is simply because most snakes continue to eat small items throughout life, incorporating larger items into their diet as they grow (reviewed in Arnold 1993). Thus, the maximum slope of this allometry will be approximately isometric (i.e. = 1) but the average slope is likely to show strong positive allometry (i.e. >1) as snake head size increases faster than average prey size. The implications of this assumption and likely bias are detailed below (Discussion).

⎛ δ(toad size) ⎞ 4. The allometry of toad size against toad toxicity ⎜ ⎟ . ⎝ δ(toad toxicity )⎠ 76

Most toxin in the skin of toads is stored in the large parotoid glands located above the shoulders (Meyer and Linde 1971). Thus, we can use the size of the parotoids as an index of the amount of toxin carried by a toad. The allometry of parotoid size should offer a reasonable surrogate measure of the allometry of toad toxicity.

I measured all toads in the collection of the Queensland Museum (QM) to obtain a broad overview of the allometry of toad toxicity across the majority of their Australian distribution. All toads were measured for snout-ischium length (SIL), head width (measured across the head at the tympanum), parotoid length (from above and behind the eye to the most distal point), and parotoid width (back of shoulders to most distal point). Both head width and SIL were used as size variables; SIL because it is a common measure of size and head width because this is the measure that influences the ingestibility of a toad by a snake. Morphometric variables were log-transformed prior to analysis. I used

Reduced Major Axis (RMA) regression to calculate slope coefficients (null slope

= 1 in all cases).

⎛ δ(snake resistance)⎞ 5. Combined allometry of snake resistance on toad toxicity ⎜ ⎟ . ⎝ δ(toad toxicity ) ⎠

Although I identified five different allometric relationships relevant to the overall relationship between snake body size and vulnerability to toad toxin

(above), my empirical data revealed that one of these did not depart from isometry (see below). This was snake resistance to toad toxin, which was 77 isometric with mass within and among species. Additionally, we have assumed that the relationship between snake head width and toad size is isometric (above). Thus to examine deviations from isometry, we only need to consider the effects of the allometry of (1) Snake mass on snake length, (2) snake length on snake head size and (3) toad size on toxicity. That is (see Figure 1):

δ(snake resistance) δ(snake mass) δ(snake length) δ(toad size) ∝ × × δ(toad toxicity) δ(snake length) δ(snake head size) δ(toad toxicity )

The product of the null values for each of these allometries produces a null value for this allometry equal to three. For each of these component allometries, I generated an estimate of the slope and the standard error around that slope. I used a randomisation procedure to arrive at a value of the product of these three estimates and the associated error. To do this I drew 10,000 random samples from each one of the distributions described by the allometry estimate and its standard error. This resulted in 10,000 sets of three samples for which I calculated the product. The resulting distribution estimated both the product and the standard error (the standard deviation of my sample distribution) around my estimate. This analysis was conducted in Excel using a purpose-written procedure. 78

8 7.5 6

7 7 5

6 6.5 4 5 6 Ln(snake length) Ln(toad length) Ln(snake mass) 3 4 5.5

3 5 2 3 4 6 7 5 6 7 8 9 2.5 3.5 6.5 7.5 Ln(snake length) Ln(snake head length) Ln(toad parotoid gland length)

Null slope 3.00 x 1.00 x 1.00 = 3.00

RMA slope 3.05 x 1.64 x 0.84 = 4.20

Figure 1. Schematic representation of the calculation of null and observed slopes, and the final allometry - representing the change in snake resistance with changes in toad size. Data shown are for the red-bellied blacksnake (Pseudechis porphyriacus). Dotted lines represent the null slope in all instances and the solid line represents the RMA regression slope. 68 79

RESULTS

1. The allometry of snake resistance to toad toxin

⎛ δ(snake resistance) δ(snake resistance) δ(snake mass) ⎞ ⎜ = × ⎟ . ⎝ δ(snake length) δ(snake mass) δ(snake length)⎠

I found no significant correlation between mass-adjusted toxin resistance and mass for either snake species that I tested (Table 1). Because there was no significant correlation, ordinary least squares regression was used to estimate slopes (RMA regression on uncorrelated variables will always yield a slope of approximately one and thus give a misleading impression of a relationship

(Sokal and Rohlf 1995)). Therefore, for further analysis I have assumed that resistance to toxin scales isometrically with body mass for all species.

Body mass, however, deviated from isometry in three of the four species that I tested (Table 2). Both the colubrids exhibited positive allometry for mass whereas one of the elapids (Hemiaspis) showed strong negative allometry.

Pseudechis showed no evidence of a deviation from isometry. I was unable to obtain data for Tropidechis and in further calculations I assume isometry for mass in this species, noting that this assumption does have implications on further calculations for this species.

⎛ δ(snake length) ⎞ 2. The allometry of snake length against head size ⎜ ⎟ . ⎝ δ(snake head size)⎠

All five snake species exhibited strong positive allometry for body size

(SVL) against head size variables. This relationship was true irrespective of the 80 head size variable considered (Table 3). Thus, head size relative to body size decreases in larger specimens within each of these species (i.e., body size increases disproportionately with an increase in head size). Thus, maximal ingestible prey size relative to snake body size should decline in larger snakes, even though maximum prey size increases in absolute terms. Hence, all else being equal, larger snakes would be less capable of ingesting a fatal dose of toad toxin. The interspecific comparison showed the same pattern as seen within the species I tested; that is, smaller species had relatively larger heads and thus, the relationship between body size and head size showed positive allometry, although of a lesser magnitude to that of the intraspecific comparison (Table 3). 81

Table 1. Results of the regression of toxin resistance scores against body

Species n OLS Coefficient s.e. r2 p Elapidae Pseudechis 30 0.012 0.0081 0.079 0.13 Colubridae Tropidonophis 24 -0.051 0.0793 0.018 0.53 mass for two species of Australian snake.

“OLS Coefficient” refers to the ordinary least squares regression coefficient. Neither regression yielded a statistically significant coefficient.

Table 2. RMA regressions of snake mass on snake length.

Snake Mass/SVL n Coefficient s.e. r2 Allometry Elapidae Pseudechis 65 3.05 0.084 0.95 0 Hemiaspis 48 2.57 0.091 0.94 --- Colubridae Dendrelaphis 14 3.71 0.325 0.9 + Tropidonophis 104 3.20 0.075 0.94 ++

The null value in this case is three. + or – denote positive and negative allometry respectively, 0 represents no significant deviation from isometry. The number of symbols in each case indicates the statistical significance of each deviation from isometry – one symbol represents p<0.05, two symbols represents p<0.01 and three symbols indicates p<0.001. 82

Table 3. Results of RMA regressions of SVL (y-axis) against head size variables (x-axis) within five species of Australian snake and across 30 species of Australian snake.

Head Length Head Width Species n Coefficient s.e. r2 Allometry Coefficient s.e. r2 Allometry Elapidae Hemiaspis 156 1.53 0.032 0.91 +++ 1.61 0.042 0.89 +++ Pseudechis 107 1.64 0.042 0.93 +++ 1.67 0.04 0.93 +++ Tropidechis 135 1.29 0.028 0.94 +++ 1.56 0.042 0.9 +++

Colubridae Dendrelaphis 275 1.44 0.024 0.92 +++ 1.63 0.037 0.86 +++ Tropidonophis 143 1.54 0.031 0.94 +++ 1.77 0.045 0.91 +++

Across Species 30 1.16 0.087 0.84 +

Allometry coefficients (slope), standard errors and r2 values are given. The ‘Allometry’ column represents the direction and significance of any deviation from isometry. + denotes positive allometry, 0 represents no significant deviation from isometry. The number of symbols in each case indicates the statistical significance of each deviation from isometry – one symbol represents p<0.05, two symbols represents p<0.01 and three symbols indicates p<0.001.

72 83

⎛ δ(toad size) ⎞ 3. The allometry of toad size against toad toxicity ⎜ ⎟ . ⎝ δ(toad toxicity )⎠

Measurements of 157 preserved toads revealed significant negative allometry for body size against parotoid size (Table 4). This negative allometry of body size was seen regardless of whether parotoid length or parotoid width was used as the indicator of toad parotoid size and whether SIL or head width was used as an indicator of toad size (Table 4). Thus, because parotoid size increases faster than body size, toads become disproportionately more toxic with increasing body size. This result is not simply the result of changes in skin morphology with age, as parotoid weight also increases disproportionately with skin weight (Phillips, unpub. data).

⎛ δ(snake resistance)⎞ 4. Combined allometry of snake resistance on toad toxicity ⎜ ⎟ ⎝ δ(toad toxicity ) ⎠

Because resistance to toxin scales isometrically with snake body size

(above), we can assess the allometry of selection by toads on snakes based simply on the remaining three allometric coefficients: Body mass on body length in snakes, head-width on parotoid size in toads, and body length on head size in snakes (see Figure 1). Multiplying these coefficients yields a net allometry that can be compared to a null value of three, to assess the overall allometry of snake resistance to toads. These overall allometric coefficients were significantly greater than 3 in all species tested (Table 5). This result reflects the fact that relative head sizes (and hence, maximal ingestible prey sizes relative to predator body size) declined with increasing body size, and did 84 so at a faster rate than toad toxicity increased with toad body size. In contrast, the equivalent interspecific analysis did not deviate from isometry (Table 5), suggesting that vulnerability to toad ingestion was not affected by a species' mean adult body size. 85

Table 4. Reduced Major Axis coefficients for regressions of head width and snout-ischium length (SIL) against parotoid size for cane toads from throughout Queensland.

Snout-Ischium Length Head Width n Coefficient s.e. r2 allometry Coefficient s.e. r2 allometry Parotoid Length 157 0.84 0.010 0.98 --- 0.83 0.009 0.98 --- Parotoid Width 157 0.79 0.010 0.97 --- 0.81 0.011 0.98 ---

- denotes negative allometry, 0 represents no significant deviation from isometry. The number of symbols in each case indicates the statistical significance of each deviation from isometry – one symbol represents p<0.05, two symbols represents p<0.01 and three symbols indicates p<0.001.

74 86

Table 5. The allometry of the relative toxicity of toads intraspecifically (within each of for seven species of Australian snakes) and interspecifically (across 30 species of Australian snakes).

Snake mass/lengthSnake length/head size Toad size/toxicity Relative Resistance of Snakes Species Coefficient s.e. Coefficient s.e. Coefficient s.e. Coefficient s.e. Allometry Elapidae Hemiaspis 2.57 0.091 1.53 0.032 0.84 0.010 3.31 0.141 + Pseudechis 3.05 0.084 1.64 0.042 0.84 0.010 4.21 0.167 +++ Tropidechis 3# - 1.29 0.028 0.84 0.010 3.25 0.027 +++*

Colubridae Dendrelaphis 3.71 0.325 1.44 0.024 0.84 0.010 4.48 0.408 +++ Tropidonophis 3.20 0.075 1.54 0.031 0.84 0.010 4.14 0.137 +++

Across Species 3# - 1.16 0.087 0.84 0.010 2.92 0.074 0*

Relative resistance allometry was calculated from the product of RMA coefficients of snake mass on length, snake body size on head size and toad body size on parotoid size in toads (from tables 2-4). See text for explanation. The null value for relative resistance allometry thus calculated is three. The last column represents the direction and significance of allometry. + denotes positive allometry, 0 represents isometry. The number of symbols in each case indicates the significance of each deviation from isometry – one symbol represents p<0.05, two symbols represents p<0.01 and three symbols indicates p<0.001. Allometries indicated with an asterix are missing the error variance of the mass/length regression and so the standard error has been underestimated. # represents an assumed value. 75 87

DISCUSSION

Cane toads have had, and will continue to have, a massive impact on

Australia’s native snakes (Phillips et al. 2003). More than 30% of Australia’s terrestrial snake fauna is at risk and many snakes are likely to die if they ingest a moderate-sized cane toad. My results indicate that the body size of a snake plays a significant role in determining the likelihood that it will ingest a fatal dose of toad toxin. Because a snake's vulnerability often does not scale isometrically with its body size, the ecological impact of toads is likely to fall disproportionately upon certain body sizes, and thus upon certain age and/or sex classes within the predator population. Such age-selective or sex-selective mortality patterns may have substantial implications for the nature and magnitude of ecological effects imposed by invading toads. For the species tested here, small individuals are predicted to be the most heavily impacted by toads. In my interspecific comparisons, however, the two major allometries cancel each other out, such that the impact of toads is likely to be independent of mean adult body size of a species. Overall, my analyses suggest that the impact of toads will not be influenced by the average body size of a species, but will be greater on smaller animals within a given predator population.

One of the allometries critical to this conclusion was assumed rather than measured: the rate at which snake head size increases with prey size. The assumption of isometry in this case is highly conservative, because in practice, many snakes continue to take small prey items as they grow (Arnold 1993). The true value of this allometry will thus be greater than 1.0 and probably 88 substantially so. If we allow the coefficient of snake head size on toad size to be greater than one (as would be the case if a snake continued to eat small prey items as it grew) the final calculation of snake resistance on toad toxicity would have an even stronger positive allometry. The assumption in this case has thus led to a very conservative bias.

It is important to note some other caveats associated with my conclusions.

1. My data are derived from very broad-scale samples, and may have

combined areas with locally heterogeneous allometries. Intensive

sampling would be needed to assess this possibility.

2. I have relied on museum samples and have thus used many individuals

to span the size range within a species. We cannot be sure that such

patterns (e.g. in relative head size) are simply the result of individual

growth trajectories unless we follow single individuals through

ontogeny (Gould 1966). The patterns observed here could be the

consequence of selection through ontogeny. In fact, it is possible that the

patterns we observe here are a consequence of selection imposed by the

presence of toads – almost all the snakes in the Queensland museum

have come from toad-exposed localities. Further research comparing

snakes from toad-exposed and toad-naïve areas would be necessary to

address this possibility.

3. Many snake species exhibit ontogenetic shifts in prey preference, such

that certain size classes within a population consume a higher proportion 89

of anurans than do other size classes. For example, the Australian

snakes Pseudonaja textilis and Boiga irregularis display an ontogenetic shift

from ectothermic to endothermic prey (Savidge 1988; Shine 1989; Shine

1991c). Such a shift would reinforce the effects of the positive allometry

for snake resistance shown here: that is, small snakes face stronger

selection from toads.

4. Gape-limitation sets an upper but not lower limit to prey sizes, so that

larger snakes often consume prey that are much smaller than their

maximum ingestible prey size (Miller and Mushinsky 1990; Shine 1991d).

Larger snakes thus have a wider choice of prey options and smaller

conspecifics will tend to eat meals that are closer to their maximum

ingestible prey size. Again, this effect will increase the effect of the

resistance on toxicity allometry described here.

5. The measure of toad toxicity used here assumes that the toxicity of toad

toxin does not change through ontogeny. No published information

exists on this issue, and I am forced to assume that toxin composition

remains approximately the same throughout a toad’s lifetime.

Nevertheless, my analyses suggest that selection in this novel predator- prey system is not distributed equally by size class within a predator species.

Specifically, selection is likely to be more intense against smaller (younger) individuals within a population. Such a bias may influence rates of recruitment and mortality, and may eventually lead to a slow population decline because 90 large adults survive but there is minimal recruitment to replace them.

Additionally, differential selection against smaller individuals may translate into a significant sex bias in vulnerability. Most anuran-eating Australian snakes exhibit sexual size dimorphism, with the male up to 22% larger than the female in some species and up to 32% smaller in others (Shine 1994).

Obviously, sex biased mortality also can have strong implications for the demographic stability and long-term viability of a population. More generally, it is important to consider the implications of allometry when considering interactions between species. Whether the interaction is being studied from the perspective of co-evolution, the impact of invasive species or competition ecology, the allometry of relevant interacting traits will influence the intensity of the interaction throughout an organism’s lifetime.

ACKNOWLEDGEMENTS

I would like to thank Greg Brown and the staff at Beatrice Hill research station for their generous hospitality. Ian Jenkins and Virginia McGrath provided logistical support and help in the field for which I am very grateful. I would also like to thank Andrew Amey, Heather Janetski and Patrick Couper

(at the QM) for access to specimens and assistance. Richard Shine, Daniel

Warner, Greg Brown and Jon Webb made helpful comments on a previous draft. Funding for this work was provided by grants from the Royal Zoological

Society of NSW, the Linnean Society of NSW, The Norman Wettenhall 91

Foundation, the University of Sydney and a grant from the Australian Research

Council.

92

CHAPTER 4

ADAPTING TO AN INVASIVE SPECIES: TOXIC

CANE TOADS INDUCE MORPHOLOGICAL

CHANGE IN AUSTRALIAN SNAKES

93

ABSTRACT

The arrival of invasive species can devastate natural ecosystems, but the long- term effects of invasion are less clear. If native organisms can adapt to the presence of the invader, the severity of impact will decline with time. In Australia, invasive cane toads (Bufo marinus) are highly toxic to most snakes that attempt to eat them.

Because snakes are gape-limited predators with strong negative allometry for head size, maximum relative prey mass (and thus, the probability of eating a toad large enough to be fatal) decreases with an increase in snake body size (small snakes eat larger prey relative to their own body mass) and increases with relatively larger head size (snakes with relatively larger heads can consume larger prey). Thus, the arrival of toads should exert selection on snake morphology, favouring an increase in mean body size and a decrease in relative head size. I tested these predictions with data from preserved museum specimens of four species of Australian snakes, collected over a period of > 80 years. GIS layers provided data on the duration of toad-exposure for each snake population, as well as environmental variables (latitude, precipitation, temperature). I used a model-selection approach to assess whether or not snake morphology has changed through time due to exposure to toads. As predicted, two toad-vulnerable species

(Pseudechis porphyriacus and Dendrelaphis punctulatus) showed a steady reduction in gape size and a steady increase in minimum body length with time since exposure to toads. In contrast, two species at low risk from toads (Hemiaspis signata and Tropidonophis mairii) showed no consistent change in these morphological traits as a function of the duration of toad exposure. These results provide strong evidence of adaptive changes in native predators as a result of the invasion of toxic prey. 94

INTRODUCTION

Human-induced environmental change is the greatest threat to global biodiversity. Such processes include global climate change, invasive species, habitat removal, over-harvesting, and altered biogeochemical cycles (Chapin et al. 2000; Diamond 1989; Novacek and Cleland 2001). These changes have caused many extinctions (local and global) and will lead to many more, but whenever the impact is non-random (i.e. selective), there is the potential for adaptive evolution. Under the right circumstances, adaptive evolution can happen very rapidly in wild populations. Such “contemporary evolution”

(sensu Stockwell et al. 2003) occurs as a consequence of selection during natural events (e.g. Grant and Grant 2002; Higgie et al. 2000; Reznick et al. 1996).

Importantly however, it has also been documented from “unnatural” (human- mediated) events. The classic example of industrial melanism in peppered moths is the most celebrated case (Grant 1999; Kettlewell 1973), however there is also clear evidence of adaptive evolution in populations as a consequence of overfishing (Olsen et al. 2004), global warming (Bradshaw and Holzapfel 2001) and heavy-metal pollution (Macnair 1987).

These studies highlight the importance of examining the potential for adaptive change in impacted populations. Doing so can clarify both the nature of the impact and the response of the affected population. Clearly, a population exhibiting an adaptive response is more likely to persist in the face of an environmental change than one that fails to adapt. Invasive species are of particular interest in this respect, because they constitute a major threat to 95 global biodiversity (Diamond 1989; IUCN 2001; Williamson 1999b). Although invasive species have caused extinctions (e.g. Ogutu-Ohwayo 1999), they may also exert non-random selection upon impacted species such that the native organisms can adapt to the presence of the invader. Although much evolutionary research has been directed towards invasive species themselves and how they adapt to new environments, much less research has been conducted on counter-adaptations by native species (D'Antonio and Kark 2002;

Lee 2002).

Many species of Australian snake have been severely impacted by the invasion of highly toxic cane toads (Bufo marinus), a conservation problem that also offers an ideal situation to explore the possibility of an adaptive response by natives to an invader. Cane toads were introduced into Australia in 1935.

Since then they have spread throughout large areas of Queensland and have entered the Northern Territory and New South Wales, currently occupying a range of approximately 1 million square kilometres (Lever 2001). The ecological impact of toads on the native fauna has been poorly elucidated, mainly due to logistical difficulties and a lack of baseline data for comparison

(van Dam et al. 2002). Nevertheless, there is a clear inference that the invasion of the toad has had a massive impact on species of Australian snakes. Toads are highly toxic and most Australian snakes attempting to eat toads will die. A recent study suggests that 49 species of snake are potentially impacted by the toad and that the majority of these species are poorly equipped to deal with a likely dose of toad toxin (Phillips et al. 2003). 96

Snakes however, are gape-limited predators and thus their ability to poison themselves by consuming a toad depends upon their head size relative to their body mass. Thus, within any given population, a snake with a small head relative to its body mass will be at less risk of ingesting a toad large enough to kill it, than will a conspecific capable of ingesting a relatively larger toad. At an intraspecific level, two major factors influence the size of a snake’s head relative to body mass: the snake’s absolute size (because smaller individuals have relatively larger heads, as is generally true in most kinds of organisms: Calder 1984) and relative head size (because even at the same body length, some individuals will have larger heads than will others). Thus, the risk of a snake consuming a toad large enough to be fatal will depend upon snake body size and relative head size (Chapter 3). Accordingly, we can expect that the arrival of toads will impose selection on the morphology of snakes, favouring individuals with larger-than-average body sizes and smaller-than- average relative head sizes.

In this study, I examine morphological variation in four species of

Australian snakes. Two of these taxa (one colubrid, one elapid) are predicted to face little to no impact from toads, either because they are too small to ingest a fatal dose or because they have high physiological resistance to toad toxins.

The other two taxa (again, one colubrid and one elapid) are predicted to be much more vulnerable to toad invasion. I examine variation in body size and relative head size with reference to environmental variables and time since exposure to toads. I predict that mean body sizes and/or relative head sizes 97 will have changed through time since toad arrival in the toad-vulnerable species, but not in the other taxa.

98

METHODS

Study species and collection of morphological data

To ensure phylogenetic independence, I selected distantly related taxa within each of my two categories (vulnerable and non-vulnerable species). All four study species feed primarily or exclusively on anurans, and are widely distributed through the parts of Queensland invaded by cane toads since 1935.

The two highly vulnerable taxa comprised red-bellied blacksnakes (Pseudechis porphyriacus, Elapidae, n = 99) and green treesnakes (Dendrelaphis punctulatus,

Colubridae, n = 242). Like most Australian snake species, these snakes are highly susceptible to toad toxins, and will die if they ingest even a relatively small toad. To consume a fatal dose of toxin, Pseudechis needs only to consume a toad whose head width is 43% of its own and Dendrelaphis needs only consume a toad whose head width is 49% of its own (Chapter 1). The two relatively non-vulnerable species were swampsnakes (Hemiaspis signata,

Elapidae, n = 158) and keelbacks (Tropidonophis mairii, Colubridae, n = 124).

These two taxa are less likely to be severely impacted by cane toads. Hemiaspis is a relatively small species with an unusually small head, and hence consumes only very small anurans even relative to its own body mass; to consume a fatal dose it needs to consume a toad that is 108% of its head-width (Chapter 1). In contrast, Tropidonophis is a larger species with a normal-sized head, but is unusual in displaying a high physiological tolerance for toad toxins; to consume a fatal dose, this species needs to consume a toad that is 185% of its 99 head-width and so feeds readily on toads without severe ill-effects (Covacevich and Archer 1975; Chapters 1 and 2).

I collected morphological data from preserved specimens held at the

Queensland Museum. Toads have populated more than 60% of Queensland and all sampled animals came from areas where toads have colonised. I selected these taxa based on their wide phylogenetic separation, their abundance, and their differential vulnerability to toads. Each specimen was measured for snout-vent length (SVL), head length (HL, base of jaw to tip of snout), head width, (HW, across the head at the junction of supraoculars and parietals) and gape width (GW, across head at last supralabial). Snout-vent length was measured to the nearest centimetre with a flexible tape, and all head measurements were taken to the nearest 0.1mm using dial callipers. Data on the date and locality of collection for each specimen were taken from museum registers.

Collection of climatic data

To minimise the chance of a spurious correlation as a consequence of spatial autocorrelation and to increase my chance of detecting an effect by reducing error variances, I included the effect of climate and latitude on snake morphology in my analyses. I derived two climatic layers for Australia using the program ANUCLIM (Hutchinson et al. 1999) and a digital elevation model of Australia with 5km grid cells. Snake locality data were laid over the resultant climate grids in ARCVIEW. The climatic data for each locality were 100 extracted using the ARCVIEW extension BIOCLIMav (Moussalli 2003). I used two climatic variables that are likely to influence snake morphology: annual mean temperature (AMT) and annual precipitation (AP).

Collection of data on duration of exposure to toads

More than 2000 records of toad locality and date were available from the

Queensland Museum and from the dataset collected by Floyd et al. (1981).

Sabath et al. (1981) and Easteal et al. (1985) used the latter dataset to map the spread of toads in Australia, however the results were hand drawn maps of the toad distribution at five yearly intervals. Improvements in mapping tools since then (i.e. GIS) allowed me to create a single digital map of far greater accuracy, which can be used to provide information on the toad expansion at yearly intervals.

I used linear interpolation of locality dates in ARCVIEW to derive a layer describing the arrival date of toads. To do this I plotted toad locality data and identified the earliest record of toads at each site by cumulatively stepping through the dataset at two-yearly intervals beginning at 1935 (the year that toads were released in Queensland). Minimum area polygons were drawn around records selected at each step and records with a later date inside each of these polygons were deleted. Following this process I used a linear spline to create a surface describing the timing of toad arrival throughout Queensland.

The resulting surface is shown in Figure 1. 101

Following the derivation of this surface, snake locality records were plotted and the year of toad arrival at each site was extracted using a spatial join. For each measured snake I subtracted the year of toad arrival (from the

GIS layer) from the collection year (from the Queensland Museum database) to yield exposure time (ET) – that is, the number of years a population of snakes had been exposed to toads at the time a snake was collected. Negative values for ET (populations that were toad-naïve at the time of collection) were converted to zero values. 102

Figure 1. GIS layer describing the timing of the cane toad invasion in Queensland, Australia. The extreme western edge of the distribution follows the extent of distribution records in Queensland and may not accurately reflect the actual invasion extent. Data from Floyd et al. (1981) and the specimen register of the Queensland Museum. 103

Data analysis

My primary interest lay in determining the effect of the presence of toads on body size and relative head size in snakes. However, other variables doubtless also influence snake morphology and thus, I needed to incorporate them into the analyses to reduce spurious correlations and so that I could focus on the residual variance – that potentially explicable by the time since toad arrival. I predicted that latitude, annual mean temperature, and annual precipitation may all influence snake morphology and so I included these, along with exposure time, as variables in a multiple regression. However, climatic and latitudinal variables were correlated to varying extents so for each species I calculated the first two principal components of climatic and latitudinal variables (PC1 and PC2) and used these as independent variables in my analysis. Two analyses were run for each species. The first used snake snout-vent length (SVL) as the dependent variable and the second used snake head-size. Head size (HS) was calculated as the first principal component of the three head size variables I measured (HL, HW and GW). The multiple regression for snake head-size also included snake body size (SVL) as a fixed independent variable as we are only interested in changes in relative head size.

In all cases, correlations between independent variables in the multiple regression were low. I log-transformed all morphometric variables and the exposure time variable was mean-centred ( y'= y − y ) prior to analysis (principal components are already mean centred). Mean-centring (such that the new mean is zero) ensures that estimated coefficients are informative even in the 104 presence of interactions; this method also reduces colinearity between variables and their interaction terms (Jaccard and Turrisi 2003).

With three independent, non-fixed variables I had 7 combinations of primary variables that could produce a model (ignoring interaction terms).

Because I had no a priori knowledge about how each variable would affect snake morphology and because the total number of models was small, I ran each of these combinations as a full model and deleted interaction terms if p- values indicated they were not significant (i.e., p > 0.05). To make model exploration and interpretation tractable, I only considered first-order interactions. For each combination of primary variables I thus derived the most parsimonious reduced model and I calculated the Akaike information criterion

(AIC) value and Akaike weight (wi) for this model. I collected the best set of models for each species and each independent variable based upon these AIC values, with models <2 units from the best model (i.e. Δi < 2) retained within the best set (Burnham and Anderson 2001). All statistical analysis was performed in JMP (v5).

Some of the models thus selected contained interaction terms. My primary interest was whether exposure time to toads was an important influence on snake morphology and, if so, the direction of the effect. The presence of interaction terms complicates the interpretation of main effects because the partial coefficient for the main effect of interest depends on the values of other variables. Mean-centring causes the main effect coefficient to be calculated for the mean value of interacting variables. However, in all models 105 with interaction terms affecting exposure time, I also calculated a range of coefficients for exposure time using values for the interacting variables that were two standard deviations above and below their mean.

RESULTS

Principal components

Principal components of climate and latitude for each species are shown in Table 1. Although results varied between specific datasets, PC1 appears to capture most of the variation due to latitude with PC2 accounting for most of the residual variation in precipitation and temperature. PC1 and PC2 accounted for between 88-97% of the variance in the three input variables.

Snake Body Size (SVL)

My models of snout-vent length suggest that, in most species, this trait was influenced by all three of the independent variables that were used (Table

2). At least one independent variable had a significant effect on snake body size for all species examined except Tropidonophis. Models for this species exhibited low Akaike weights and very low r2 values indicating that all the independent variables explained negligible variance in body size for this species. 106

Time since exposure to toads appeared in the best model set for all species but only indicated a significant effect in Pseudechis and Dendrelaphis. In both these cases, the effect of ET on snake body size was positive indicating that these species – the two most vulnerable to toads – increase in average size with increasing exposure time. An interaction between exposure time and PC2 acts as a modifier to the partial coefficient of ET for Pseudechis but within the bounds defined by 0 ± 2σPC2 the partial coefficient did not change sign, remaining positive.

Across all four species, the predicted impact from toads (i.e. the relative size of toad that each species would need to consume to ingest a lethal dose, see methods) was highly negatively correlated with the mean coefficient for the effect of ET on SVL (r = -0.89, n = 4, n.s.; or r = -1.00, n = 3, p = 0.005 if Pseudechis is excluded because of the interaction between ET and PC1; Figure 2). This indicates that the predicted level of impact from toads predicts the rate of response in body size.

Snake Relative Head-Size

Exposure time and PC1 contributed significantly to variation in snake relative head-size (Table 3). Models for Tropidonophis again showed very low

Akaike weights (none of the eight models could be considered notably better than the others) suggesting that none of the independent variables explained much of the variation in relative head-size in this species. In the remaining 107 three species, ET had a significant negative effect on relative head-size. For

Hemiaspis however, this negative effect became positive at low values of PC1, as a consequence of the interaction between these terms. Thus we are left with an unequivocally negative effect for only two species, Pseudechis and Dendrelaphis, the two facing the highest impact from toads. For these species, relative head size decreases with time since exposure to toads.

Across all species, the predicted impact from toads was highly correlated with the coefficient for the effect of ET on relative head-size (r = 0.73, n = 4, n.s.; or r = 0.99, n = 3, p = 0.055 if Hemiaspis is excluded due to the interaction between ET and PC2; Figure 2) indicating that the relative impact of toads also affects the rate of response in relative head-size. 108

Table 1. Description of the first two principal components of climatic and latitudinal variables for each snake species.

Species Principal Eigenvalue Cumulative Eigenvectors component percent AMT Aprecip DecLat

Hemiaspis PC1 1.88 62.69 0.49 0.57 -0.66 PC2 0.81 89.83 0.79 -0.60 0.07

Pseudechis PC1 2.14 71.35 0.55 0.58 -0.60 PC2 0.53 88.86 0.80 -0.58 0.17

Dendrelaphis PC1 2.03 67.63 0.66 0.32 -0.68 PC2 0.89 97.15 -0.28 0.94 0.18

Tropidonophis PC1 1.99 66.17 0.66 0.30 -0.69 PC2 0.92 96.71 -0.29 0.95 0.13

Principal components were constructed from three raw variables; Annual Mean Temperature (AMT), Annual Precipitation (APrecip) and Decimal Latitude (DecLat). Cumulative percent describes the cumulative percentage of the total variance captured by the principal components.

93 109

Table 2: Parameter estimates for best model sets for multiple regression analyses of body size (SVL) in four species of Australian snake.

2 Species r ² i w i Intercept ET ET Range PC1 PC2 Interactions Hemiaspis 0.054 0.00 0.46 0.002 -0.089 0.058 1.38 0.23 0.002 -0.016 -0.090 0.054 1.96 0.17 0.003 0.002 -0.029 -0.092

Pseudechis 0.196 0 0.49 0.007 0.016 0.004-0.028 -0.092 ET*PC2, 0.0083 0.153 0.85 0.32 0.001 0.014

Dendrelaphis 0.077 0.00 0.64 -0.003 0.006 0.048 0.077 1.95 0.24 -0.003 0.006 0.048 0.007

Tropidonophis 0.014 0.00 0.27 0.007 -0.042 0.023 0.85 0.18 0.006 -0.004 -0.049 0.006 0.85 0.18 0.005 -0.003 0.000 1.53 0.13 0.006 0.008 0.014 1.97 0.10 0.007 -0.042 0.008

Three independent variables were used: time since exposure to toads (ET) and two principal components (PC1 and PC2) incorporating data on latitude, annual mean temperature and annual precipitation. Parameter estimates significantly different from zero are shown in bold. wi is the Akaike weight of each model and Δi refers to the change in AIC value from the best model.

94 110

Table 3: Parameter estimates for best model sets for multiple regression analyses of head size in four species of Australian snake.

2 Species r ² i w i Intercept SVL ET ET Range PC1 PC2 Interactions Hemiaspis 0.9402 0.00 0.71 -0.295 5.267 -0.006 -0.017-0.006 0.030 ET*PC1, 0.0042 0.9402 1.97 0.26 -0.293 5.263 -0.006 -0.016-0.006 0.031 -0.007 ET*PC1, 0.0041

Pseudechis 0.9569 0.00 0.26 0.040 3.040 -0.005 -0.081 0.9556 0.46 0.20 0.041 3.049 -0.005 0.9575 0.83 0.17 0.033 3.034 -0.007 0.036 -0.087 0.956 1.69 0.11 0.039 3.045 -0.007 0.030 0.9548 1.91 0.10 0.042 -0.079

Dendrelaphis 0.9233 0.00 0.77 0.016 3.937 -0.005 -0.194 0.012 PC1*PC2, 0.0526

Tropidonophis 0.9485 0.00 0.19 -0.049 3.201 0.043 0.034 PC1*PC2, -0.0655 0.9455 0.07 0.18 -0.051 3.175 0.9464 0.35 0.16 -0.052 3.187 0.035 0.9491 0.56 0.14 -0.033 3.193 0.002 -0.013-0.017 0.046 0.007 ET*PC1, -0.0056 0.9477 1.55 0.09 -0.001 3.176 0.001 -0.012-0.016 0.012 ET*PC1, -0.0050 0.9467 1.67 0.08 -0.050 3.193 0.002 0.039 0.9457 1.76 0.08 -0.049 3.178 0.001 0.9456 1.99 0.07 -0.051 3.175 0.011

Snake body size (SVL) is included as a fixed independent variable as we are only concerned with relative head size. Three independent variables were used: time since exposure to toads (ET) and two principal components (PC1 and PC2) incorporating data on latitude, annual mean temperature and annual precipitation. Parameter estimates significantly different from zero are shown in bold. wi is the Akaike weight of each model and Δi refers to the change in AIC value from the best model. 95 96

0.02

0.015 Pseudechis

0.01

Dendralaphis 0.005

Hemiaspis 0 Body size ET parameter

Tropidonophis -0.005 0 50 100 150 200

0.002 Tropidonophis

0

-0.002

-0.004

Dendralaphis

-0.006 Pseudechis Hemiaspis Head size ET parameter

-0.008 0 50 100 150 200

Predicted impact

Figure 2. Parameter estimates describing the rate of change in body size and head size for each snake species, plotted against the predicted impact from toads (from Chapter 1). See text for statistical tests.

97

DISCUSSION

The results of my modelling strongly support the prediction that

Australian snakes will display morphological adaptations that reduce their vulnerability to cane toads. The duration of exposure to toads was significantly associated with changes in mean body size and relative head size in the two snake species that were identified (from previous work) as being extremely vulnerable to toads. Importantly, the changes occurring since toads arrived were in the directions predicted by my hypothesis of size-dependent vulnerability (i.e., mean body sizes have increased, and relative head sizes have decreased). In contrast, the two taxa that were identified as being less vulnerable to toads, showed fewer (or no) significant changes in morphology associated with the presence of these toxic anurans. My modelling suggested that exposure to toads may influence head-size for one of these species, but the exact nature of any such effect remains obscure. There was much less ambiguity about associations between morphology and the duration of exposure to toads in the two toad-vulnerable species.

Furthermore, the rate of change in morphology as a consequence of exposure to toads appears to be linearly related to the predicted level of impact for each species. This is exactly what we would expect if the rate of response to toads was driven by the strength of selection imposed by toads, all else being equal.

In the two toad-vulnerable species (Dendrelaphis and Pseudechis), there is a significant increase in mean body size in populations sympatric with toads.

98

Because small individuals face a much higher risk of fatal poisoning by toads

(Chapter 3), these shifts through time likely result from an ongoing loss of small individuals from populations exposed to toads. At its simplest, this effect may be the product of consistently high mortality rates among juvenile snakes in each generation, such that the population structure in toad-exposed areas is shifted towards larger, older animals. This would imply strong selection against small body size, but not necessarily longer-term adaptation. Another possibility is that the presence of toads elicits a developmental response (such as increased growth rate through increased food availability) leading to fewer small individuals. The final possibility is that toads have exerted significant selection on life-history tactics of the snakes, such that populations in toad- exposed areas now produce larger (and presumably, fewer) offspring per clutch, or the young snakes (independent of changes food availability) grow more rapidly to a size at which they become less vulnerable to toads.

Similar ambiguity in interpretation also occurs with the causal processes responsible for changes in relative head size. Both the “vulnerable” species,

(Dendrelaphis and Pseudechis), as predicted, showed a significant decrease in head size associated with time since exposure to toads. This could be due either to an ongoing impact and an adaptive response to that impact or, alternatively, to developmental changes in head growth associated with dietary change subsequent to the arrival of toads. Although early studies reported that relative head sizes in snakes were not developmentally plastic with respect to temperature (Arnold and Peterson 1989; Forsman 1996), recent studies provide

99 evidence that relative head sizes in snakes can shift as a consequence of differences in mean prey size (Bonnet et al. 2001; Queraz-Regal and King 1998).

Although I have no direct data to distinguish between these two scenarios

(selection versus plasticity), my data argue against an indirect environmental effect. Because the effect of exposure time across species is strongly related to the likelihood of a species ingesting a toad large enough to kill, we can be confident that the observed effect is a consequence of a direct interaction between toads and snakes. In other words, the morphological effect is not driven by changes in prey abundance, prey size or other indirect environmental effects. Therefore, the morphological changes must be a consequence of (and probably also a response to) selection against small bodies and large heads.

If we accept that the morphological changes observed are a consequence of selection then the obvious corollary is whether this selection is resulting in evolutionary change. Although there are a number of reasons why populations might not respond to selection (Merila et al. 2001), the simplest is that there is insufficient heritable variation at the traits of interest. To be confident that the observed effect here is an evolved response to selection we would need to understand the heritability of head size and body size (particularly offspring size) in our species. While these estimates were not available, it is generally accepted that there is almost always heritable variance available in populations, particularly for life-history and morphological traits (although see Hoffmann et al. 2003 for an exception; Roff 1997). Certainly, recent work by Sinervo and

Doughty (1996) showed a very high heritability (0.62) for egg size (i.e. offspring

100 size) in a species of lizard, and egg size in birds typically shows high levels of heritability also (>0.5, Christians 2002). Morphological traits (e.g. head-size), tend to have heritabilities of around 0.4 (Roff 1997). It seems very unlikely that snake populations are not responding to this selection due to a lack of heritable variation.

Additionally, the relatively short generation time of these snake species snakes (<3 years, Shine 1978) allows more than 20 generations to have elapsed since initial exposure to toads in some areas. These facts suggest that offspring- size and head-size are likely to have had both the time and lability to exhibit evolved change. Further research on possible life-history shifts in toad-exposed predator populations would be of great interest.

The general approach outlined here, using a combination of museum time-series and spatial data, could be used to assess morphological change in any species provided that it is adequately represented in collections and the spatial timing of the change can be mapped. This highlights not only the relative ease with which impacts of and responses to environmental change can be assessed when the relevant data are available, but also the importance of museums as storehouses for specimen series that can be used to examine temporal processes.

While the snake species examined here exhibited morphological change in response to impact from toads, there is also reason to believe that other traits may be under selection simultaneously. Traits such as prey preference (the tendency of a snake to eat toads), resistance to toxin and habitat choice are all

101 likely to be under selection and may be showing similar adaptive responses.

Chapters five and six assess these possibilities.

The data presented here indicate an adaptive response by a population impacted by an invasive species. As such these results can be added to a growing list of studies suggesting rapid adaptation associated with environmental change. Furthermore, this study demonstrates adaptive change in response to impacts from an invasive species; it is one of the first studies to do so (the only other we are aware of being Kiesecker and Blaustein 1997).

Clearly, the potential for impacted populations to adapt needs to be considered when assessing long-term impacts of environmental change. Assessing the possibility and extent of an impact associated with an environmental change is a useful first step (e.g. Thomas et al. 2004), but the next logical step is to assess the potential for impacted species to adapt. Without such information, we cannot predict the long-term consequences of environmental change.

ACKNOWLEDGEMENTS

I would like to thank Patrick Couper, Andrew Amey and Heather

Janetski at the Queensland Museum for access to specimens and many cups of tea. I am also grateful for the GIS advice and technical assistance provided by

Adnan Moussalli and Michael Kearney. Earlier drafts were improved by comments from Richard Shine, Ary Hoffmann and Greg Brown. The

Australian Research Council provided funding.

102

103

CHAPTER 5

ASSESSING THE POTENTIAL FOR AN

EVOLUTIONARY RESPONSE TO RAPID

ENVIRONMENTAL CHANGE: INVASIVE

TOADS AND AN AUSTRALIAN SNAKE*

* Published as: Phillips B L, Brown G P and Shine, R, 2004. Assessing the potential for an evolutionary response to rapid environmental change: Invasive toads and an Australian snake. Evolutionary Ecology Research 6:

104

ABSTRACT

Extinctions are ultimately caused by a change in an organism’s environment.

Species that can adapt are more likely to persist indefinitely in the face of such changes.

I argue that an understanding of the factors encouraging and/or limiting the potential for adaptation is an important consideration in assessing the long-term outcomes of environmental change. Such an approach suggests a cohesive way of assessing the potential for an impact and the long-term consequences of a particular environmental change. I illustrate this approach with a case study of a native Australian snake (the keelback, Tropidonophis mairii) faced with the invasion of an extremely toxic prey item (the cane toad, Bufo marinus). I examine the likely strength of selection, the heritability of toxin resistance and the likelihood of trade-offs or pre-adaptation. I assess an internal trade-off (between toxin resistance and locomotor performance) and an external trade-off (between resistance to the toxin of toads and a native prey species,

Litoria dahlii). My analysis reveals weak selection, high heritability and no trade-offs in resistance to toad toxin, suggesting that keelbacks are capable of mounting a rapid adaptive response to invasion by the cane toad.

105

INTRODUCTION

Rapid environmental change is currently the biggest threat to global biodiversity. Such changes include global climate change, invasive species, habitat removal, altered biogeochemical cycles and others less ubiquitous

(Chapin et al. 2000; Novacek and Cleland 2001). These changes have led to many extinctions (local and global) and will lead to many more. However, environmental change is not a new phenomenon – the history of the earth reveals that many major changes have occurred in the past. Continents have moved, climate has changed and previously isolated communities have been thrust together (Morrison and Morrison 1991). What makes anthropogenic environmental change unusual is its rapidity and ubiquity (Chapin et al. 2000).

Whether a species will persist or go extinct in the face of rapid environmental change will largely depend upon the species’ ability to adapt to the change. While the question of whether or not a species will go extinct in the short term is important, whether or not it will adapt should be the ultimate concern. Adapting to the new environment is clearly the best “strategy” for long-term persistence of a species.

This perspective suggests an approach to assessing the potential long- term impact of an environmental change. Whether a species is likely to adapt to a given change will depend on several factors including:

1. The per generation strength of selection imposed by the change

(excessively high selection may lead to extinction, whereas lower levels

of selection will encourage adaptive evolution)

106

2. The ability of the population to respond to the selective force (i.e. does

the population have sufficient heritable variation at the characters under

selection?)

3. Whether the population has other constraints limiting adaptive potential

(e.g. trade-offs, long generation-time relative to the pace of change).

Invasive species offer an excellent opportunity to study the evolutionary implications of rapid environmental change for several reasons. First, the timing and spatial pattern of the invasion is often well documented. Second, invaders often have a large impact on native species and the mechanism of impact is often well understood. Finally, invasions of species can happen naturally in ecosystems and thus it is expected that some capacity to adapt to them should exist.

In this paper, I describe a case study in which I assess the likelihood of a rapid adaptive response by a native species to an invader. To do so I assess the heritability of an adaptive trait and the likelihood of trade-offs (both intrinsic and extrinsic) acting against adaptive change. This approach is logistically simpler than documenting short term impacts, and can clarify probable long- term effects of the invader on the native species.

A case study: cane toads and Australian snakes

The history of the cane toad in Australia represents an excellent example of the invasion of a dangerous prey item. Cane toads (Bufo marinus) were

107 introduced into Australia in 1935. Since then they have spread throughout large areas of Queensland and have entered the Northern Territory and New

South Wales, currently occupying a range of approximately 1 million square kilometres (Lever 2001). The exact impact of toads on the native fauna has been poorly elucidated, mainly due to logistical difficulties and a lack of baseline data for comparison (van Dam et al. 2002). Nevertheless, there is a very clear inference that the invasion of the toad has had a massive impact on species of

Australian snakes. A recent study suggests that 49 species of snake are potentially impacted by the toad and that the majority of these species are poorly equipped to deal with a likely dose of toad toxin (Phillips et al. 2003).

The main active principal of toad toxin is a class of steroid-derived compounds known as bufogenins (or bufodienolides, Chen and Kovarikova

1967), unique to toads and biochemically very different from the active peptides which constitute the main defensive secretions of Australian frogs (Daly and

Witkop 1971; Erspamer et al. 1984; Erspamer et al. 1966). Bufogenins are extremely toxic, exerting strong cardiac effects. Thus with the arrival of the cane toad, Australian snakes were faced with a novel and extremely powerful toxin in potential prey items.

Australian frog-eating snakes are thus under selective pressure to adapt to the presence of the toad. Four possible adaptive solutions are identifiable

(Brodie and Brodie 1999a):

1. Populations can increase resistance to toad toxin,

2. populations can evolve to avoid/exclude toads as a prey item,

108

3. populations can evolve modified habitat preferences (spatial and/or

temporal) such that exposure to toads is reduced and/or

4. populations can evolve shifts in morphology (particularly relative gape

width) that reduce ingestible prey size and hence, exposure to lethal

doses of toxin.

None of these solutions are mutually exclusive and the presence of toads is likely to drive populations towards all four. Whether each solution is achievable will depend upon the magnitude of response required (the strength of selection), the heritable variance for each relevant trait and the presence or absence of trade-offs or pre-adaptations. However, any one of these solutions, on its own, may allow a population to persist with toads.

This paper explores the possibility of adaptation by increased resistance to toad toxin in one species of snake (the keelback, Tropidonophis mairii). In doing so I examine the strength of selection, the heritable variation and the possibility of trade-offs or pre-adaptation as a result of co-evolution with dangerous native prey (Dahl’s aquatic frog, Litoria dahlii).

A relevant pre-adaptation to toad toxin could exist through previous, long-term exposure to other toxic amphibians. Although the effects of native frog toxins on Australian snakes are poorly known, most Australian frogs contain skin toxins, yet many are still eaten by frog-eating snakes (Greer 1997).

One exception to this generality is Litoria dahlii. This frog is extremely toxic to most sympatric snake species and is abundant in floodplains of tropical

Australia (Madsen and Shine 1994). The only snake species found capable of

109 consistently surviving the ingestion of Litoria dahlii was the keelback,

Tropidonophis mairii (Madsen and Shine 1994). This species is also extremely resistant to the toxin of the cane toad (Phillips et al. 2003).

Examining this system (T. mairii, L. dahlii and B. marinus) allows us to ask the specific question: Are keelbacks likely to adapt to the presence of toads? To assess the potential for adaptive response in toxin resistance I:

1. calculate the heritability of toxin resistance (heritability of a trait is the

proportion of the variation in the trait that is directly heritable and thus

gives an indication of the “ability” of the trait to respond to selection),

2. examine possible trade-offs (strong trade-offs may constrain adaptive

options) and

3. examine the possibility of pre-adaptation through co-evolution with

native prey (pre-adaptation potentially reduces the strength of selection).

In making these assessments and using T.mairii as a model, I am also able to address several questions of general relevance to the adaptive solutions available to Australian snakes. Firstly, Tropidonophis is highly resistant to both

L. dahlii and Bufo toxins. This raises the possibility that resistance to one toxin confers resistance to the other (i.e. pre-adaptation to Bufo through co-evolution with L. dahlii). If this is the case, we might expect a lowered impact on snakes in areas where L. dahlii is abundant due to similar selective forces imposed by both toads and L. dahlii.

110

Secondly, the relationship between resistance to the two toxins may be the reverse of that suggested above: adaptation to one toxin may reduce resistance to the other. The presence of such a trade-off would restrict adaptive options for Australian snakes faced with the selective force of the presence of toads — adapting to one dangerous prey would reduce their ability to tolerate the other dangerous prey taxon.

Thirdly, adaptation to either prey may entail decreased performance in other traits related to fitness. One such trait might be decreased locomotor performance associated with higher levels of resistance to the toxin. Brodie and

Brodie (1999b) argue that there is a strong trade-off between resistance to tetrodotoxin and locomotor performance in garter snakes (i.e., faster snakes are disproportionately affected by toxin).

111

METHODS

Toxin Extraction

I extracted toxin from freshly killed L. dahlii and B. marinus by removing the dorsal skin (front of parotoids to knees in B. marinus and back of tympanum to knees in L. dahlii) and drying it at room temperature before weighing. Dried skins were cut into pieces and placed in a blender with 10 times v/w of 40% ethanol. Skins were rehydrated for 2 – 3 h before blending. The resulting liquid was strained and then reduced to 50% of initial volume by evaporation at room temperature. These preparations were stored at 4°C between use. A control solution was created using 40% Ethanol evaporated to 50% of its original volume at room temperature.

Collection of snakes

Gravid female keelbacks were collected by hand near Humpty Doo, in the Northern Territory. Females were measured and weighed and kept in captivity until they oviposited (usually within 1 week of capture). At the time of collection, toads were not yet present in this area; the invasion front was approximately 300km south and west of Humpty Doo.

Newly laid eggs were measured and weighed and placed on a mixture of vermiculite and water in a plastic bag to incubate. Thirteen clutches were split into four incubation treatments — two hydric regimes (wet=1:1 ratio of vermiculite to water by mass, dry=2:1 ratio) were run orthogonally with two

112 temperature regimes (high variation, mean 23.4°C, variance 9.4 and low variation, mean 23.5°C, variance 5.1). The remaining 13 clutches were incubated under conditions identical to the wet substrate/low temperature variance treatment.

Toxin Resistance Assay

Resistance to toxin in newly-hatched keelback snakes was assayed using post-dose reduction in locomotor performance, a methodology modified from that of Brodie and Brodie (1990). Individuals were swum along a 2m swimming trough and were timed with an electronic stopwatch over three consecutive 50cm segments of the trough with a stopwatch. Animals were encouraged to swim by tapping them on the tail. Water temperature was maintained at 23±1oC.

All individuals were weighed before testing to the nearest 0.1 g on a digital scale. A swimming trial consisted of two consecutive laps of the trough.

This yielded six measurements of swim speed over 50cm of which only the fastest was retained. All animals (1-2 days post hatching) were initially subjected to three swim-trials one hour apart. This yielded three maximum sprint speed times which were averaged to generate the pre-dose estimate of maximum swim speed (b).

On the following day, snakes were given a specific dose of toxin or control solution. Dosing was achieved by use of a micropipette attached to a thin rubber feeding tube. The tube was inserted into the stomach to a depth of

113

5cm from the snout before toxin was expelled. Animals were observed for 1 minute following this procedure to ensure the toxin was not regurgitated. Two swim trials were then undertaken for each individual 30min and 90min post dosing. A third trial was not run due to the increased possibility of recovery by this time. Once again only the fastest speed for each trial was taken. This yielded two time measurements which were averaged to give post-dose swim speed (a). The percentage reduction in swim speed (%redn) was calculated from these times using the formula, %redn = 100 x (1-(b/a)).

Locomotor effects

Several experiments were performed. The first two were to determine whether the administration of toxin caused a reduction in swim speed. For toad toxin I was also able to assess the effect of increasing dosages on the decrement in swim speed. Due to limited toxin extract I did not assess this factor for L. dahlii toxin.

For L. dahlii toxin, two clutches of keelbacks (previously untested, single incubation treatment, 19 individuals) were used. Each clutch was split into two groups: Group 1 was given 50μL of L. dahlii toxin and group 2 was given the same volume of control solution in the manner described above. The post-dose decrement in swim speed for each individual was recorded. A one factor

ANOVA was used to determine the effects of toxin vs control on the decrement in swim speed.

114

For toad toxin, five clutches of keelbacks (previously untested, single incubation treatment, 24 individuals) were split by clutch into four groups.

Each group was given either 100μL of control solution, 25, 50 or 100μL of toad toxin. Decrement in swim speed was assessed as previously described. A one factor ANOVA was used to assess the effect of dose level on the decrement in swim speed.

115

Toxin resistance trade-offs: Litoria dahlii versus Bufo marinus

To assess the correlation between responses to toad toxin and responses to L. dahlii toxin within individual snakes I used 15 clutches (from both the mixed and single incubation treatments) and tested all 115 individuals for resistance to both toad and L. dahlii toxins. On the day following pre-dose swimming trials, neonates were tested with 50μL of toad toxin. On the fourth day following dosing with toad toxin, animals were given 50μL of L. dahlii toxin and tested in the same manner as for toad toxin.

The order of dosing was not staggered because my first priority was to develop a large dataset on toad resistance for the heritability analysis (see below). However I did conduct a small experiment to determine whether being previously tested for toad toxin affected L. dahlii resistance measures. In this experiment I split four clutches (18 individuals) into two groups. One group was tested first for resistance to toad toxin and then for resistance to L.dahlii toxin. The second group had the testing order reversed. In both groups, four days elapsed between each test. Data from the individuals tested for toad toxin resistance first was also used in the heritability analysis (see below).

Data Analysis

Heritability

I measured locomotor decrements of 167 individuals representing 24 clutches (including individuals assessed for resistance to L. dahlii toxin tradeoffs and toad toxin locomotor effects) to estimate heritability using a full-sib design.

116

Full-sib designs are unable to control for the effect of a common maternal environment, potentially inflating estimates of heritability. Of the 24 clutches used for heritability analysis, 13 had been incubated under varying conditions

(see above). This process effectively minimised the common incubation environment of clutches and allowed me to explicitly test the effect of incubation treatment on toxin resistance. Variance components under the full- sib design were estimated by restricted maximum likelihood (REML) in SPSS.

REML is the best technique for generating unbiased estimates of variance parameters with an unbalanced design (Shaw 1987). I used a jacknife approach

(iteratively removing one family) to estimate heritability and its standard error.

Both neonate mass and maternal mass may influence resistance to toad toxin.

Before calculation of heritability, these factors were removed from toxin resistance data by taking the residuals of a multiple regression of both neonate mass and maternal mass on %redn.

Locomotor Trade-offs

To assess the possibility of a trade-off between toxin resistance and locomotor performance, I regressed pre-dose speeds against post-dose speeds for 15 clutches (115 individuals) tested with both toxins. The gradient of the line in each case was assessed against null expectations under the following model:

A = mB + c

117 where A is the post-dose speed, B is the pre-dose speed, m is the effect of the toxin (null: equal to the average proportion of original speed for each toxin) and c is a constant. This method for inferring the presence of a trade-off differs from that used previously for similar data (Brodie and Brodie 1999b), with the crucial difference being in construction of the null slope. My approach is more conservative, because the null slope under my model will always be less than the null of 1 assumed by Brodie and Brodie (1999b). Furthermore, the test is made more conservative because individual mass could not be removed from the analysis without compromising the logic of calculating the null. The inclusion of mass will bias the slope upwards, resulting in conservative inference under this analysis.

A randomisation test was performed to assess the significance of any deviation of m from that expected under the null hypothesis (Manly 1991). In this case, a trade-off is evidenced by a slope lower than the null. The effect of individual mass on response to the standard dose of toxin could not be removed without sacrificing the ability to calculate a null slope. The effect of individual mass is likely to bias the observed slope in a positive direction making this a conservative test for the presence of a trade-off in this system. To determine the power of the data to detect a deviation from the null slope, the randomised distribution was advanced by successive decreases of 0.05. For each of these increments the overlap of the new distribution with non- significant values of the test distribution was calculated (β). The power was calculated as 1-β for each increment (Sokal and Rohlf 1995).

118

Toxin resistance trade-offs: Litoria dahlii versus Bufo marinus

Because both toxins were tested on each individual in this experiment, there is no need to adjust post-dose times by pre-dose times. In fact doing so would cause the measures for the two toxins to become correlated through the effect of the common pre-dose speed. Thus post-dose times (a) for each toxin for each individual were compared. Because post-dose speed is likely to be correlated with both pre-dose speed and snake mass, I removed their effect by taking the residuals of a multiple regression of pre-dose speed and snake mass on post-dose speed for each toxin. The Pearson product-moment correlation coefficient was calculated to determine the strength of the correlation between

L. dahlii and toad toxin post-dose times after correcting for pre-dose speed and snake mass. A randomisation test (Manly 1991) was utilised to test the significance of the observed correlation and to assess the power of the data to detect a correlation of various strengths. The value of the observed correlation was compared with that of 1999 randomised sets of the data. For non- significant results, power was assessed as above.

119

RESULTS

The effect of toxins

The toad skin extract amounted to 144.7mg of skin per ml of final extract.

The extract of L. dahlii skin equated to a final concentration of 119.0 mg per ml of final extract. Three snakes of 208 tested died after I gave them a dose of toxin.

Whether this was due to the toxin or simply handling stress is not discernable.

Most animals that were tested recovered over the course of 8 to 24 hours

(Chapter 2).

The mean masses of snakes did not differ significantly between treatment groups in the L. dahlii toxin experiment (F1,17=0.90 p=0.35). The dose of toxin (equating to 5.95 mg of dried L. dahlii skin) gave an average reduction in speed of 27.9% in the toxin group as opposed to a 9.4% increase in speed in the control group (F1,17=5.88, p=0.026).

A similar pattern was observed for toad toxin. The percentage reduction increased with increasing dose of toxin (Fig. 1). The difference between doses was significant overall (F3,20=16.003 p<0.0001). Fisher’s PLSD revealed significant differences between all dosage levels and the control (p<0.037 in all cases) and significant differences between each dosage level (p<0.023 in all cases) except for the 50/100μL comparison (p=0.091). 50μL of toxin gave an average reduction of 66.1% compared to a reduction of 19.5% in the control group.

120

100

75

50 % reduction 25

0 25uL 50uL 100uL Control

Treatment

Figure 1. Percentage reduction in swimming speed in neonate keelbacks across a range of doses of toad toxin. Different individuals were tested at each dosage level. Error bars represent standard errors.

121

Heritability

Multiple regression revealed significant effects of both individual

(neonatal) mass and maternal mass on toxin resistance (mass, p=0.0025; mother’s mass, p=0.0187; R2=0.058). The residuals from this analysis were used in subsequent analyses.

A total of 83 eggs hatched in the temperature/moisture experiment.

Residual toxin resistance of the hatchlings was not significantly affected by the interaction between temperature and moisture treatments (F1,79 = 1.378, p=0.24).

After removal of this factor, I detected no significant effect on residual toxin resistance of either temperature or moisture treatments (temperature, F1,80=0.94, p=0.33; moisture, F1,80=0.009, p=0.93).

Restricted maximum likelihood provided an estimate of between-clutch variance of 43.85 compared with a within-clutch variance of 148.36. This yielded an estimation of 0.456 for full-sib heritability. Jacknifing provided a standard error of 0.0488 for this heritability estimate. This high heritability does indicate partial pseudoreplication in the design of my locomotor experiments

(above), which treated siblings as independent in statistical analyses. As such these results (particularly for the experiment on the effects of L. dahlii toxin, which only used two families but for which the heritability is unknown) should be interpreted with caution.

Locomotor trade-offs

122

The average %redn across all snakes tested with 50uL of toad toxin was

62.5%. The null slope of after-speeds on before-speeds was thus 0.375. The observed slope was 0.325 and was not significantly less than the null (p=0.101).

Power analysis suggested that my data had >95% power to detect a negative deviation from the null of 0.16 and had >50% power to detect a deviation of

0.08. In comparison, the average percentage reduction across all animals tested with L. dahlii toxin was 30.8%. In this case, the observed slope of 0.4 was significantly less than the null of 0.69 (p<0.0005, Figure 2).

Toxin Trade-offs

Testing snakes for resistance to Bufo toxin before testing them for resistance to L. dahlii toxin had no significant effect on the estimate of resistance to L. dahlii toxin (F1,16=0.084, p=0.78). Including data from this experiment, a total of 115 neonate keelbacks were tested for their resistance to both L. dahlii and toad toxin. Of these, three snakes were unable to swim after dosing with toad toxin. Because I was unable to record a time for these animals they were deleted from the analysis rather than being assigned a large but arbitrary time.

Multiple regression revealed significant effects of pre-dose speed and snake mass on post-dose speed for L. dahlii toxin (pre-dose speed p<0.0001; mass p=0.0002). For toad toxin however, only pre-dose speed was significant (pre- dose speed p=0.0045; mass p=0.5689). Using the residuals from these regressions the product-moment correlation coefficient was calculated and compared with the null hypothesis of no correlation (H0:r=0, Ha: r≠0). The

123 observed correlation for the data was r=0.044 (Fig. 3). The significance of this r- value was compared with the distribution of r-values obtained from 1999 randomisations of the dataset. 31.35% of random r-values were greater than that observed, equating to a two-tailed probability of p=0.628. Because this is a non-significant result it is important to assess the power of the dataset to detect various levels of r. The power analysis suggested that the data had >95% power to detect an r-value ≥ 0.35 and >53% power to detect r-values ≥ 0.2.

124

30 A. Bufo

25 Observed Null 20

15

10

5

0

30 B. L. dahlii

25 After Speed (cm/s)

20

15

10

5

0 0 10 20 30 40

Before Speed (cm/s)

Figure 2. Scatterplot of pre-dose vs post-dose swim speeds for snakes given; A) B. marinus toxin and, B) L. dahlii toxin. Null and observed slopes are plotted. The observed slope is significantly smaller only for snakes given L. dahlii toxin p<0.0005.

125

25

20

15

10 residuals

5 Bufo

0

-5 0 1 2 3 -3 -2 -1

L. dahlii residuals

Figure 3. Scatterplot post-dose speeds for L. dahlii toxin and B. marinus toxin after correcting for pre-dose speed and body mass.

126

DISCUSSION

The effect of toxins

My results show firstly that the methodology provides a sensitive and non-lethal assay of the effect of toxin on snakes. Both toxins elicited a reduction in speed in hatchling keelbacks relative to a control dose. For toad toxin at least, this reduction strongly depended on dosage. Small differences in dosage rate caused detectable changes in post-dose speed even with relatively small sample sizes, suggesting that this methodology will be suitable for assaying variation in response to a standard dose of toxin.

A very similar methodology has been used to assay tetrodotoxin resistance in several genera of North American snakes, yielding similar results.

That is, increased doses elicited greater reduction in locomotor speed

(Motychak et al. 1999). This same pattern has also been shown for nine other species of Australian snake exposed to toad toxin (Phillips et al. 2003). While the mechanism contributing to the decrease in speed is likely different between toad toxins (action primarily cardiac), tetrodotoxin (a neurotoxin) and L. dahlii toxins (action under investigation) the basic effect appears to be simply that a sick animal swims more slowly than a healthy animal. On this basis, it seems likely that the assay will also be useful in detecting variation in resistance to most toxins.

127

Keelbacks are more resistant to toad toxin than are any of the other

Australian snakes studied to date (9 species across 3 families). It is unlikely that an individual would be able to ingest a large enough volume of cane toad to acquire a lethal dose (Phillips et al. 2003) although some instances of keelbacks apparently dying following the ingestion of a toad have been observed (Ingram and Covacevich 1990, Phillips, pers. obs.). Irrespective, individuals ingesting toads are likely to incur short-term locomotor deficits. The long-term effects of a diet of toads are unknown but there may well be serious fitness costs (Chapter

2) and, in one study, keelbacks maintained exclusively on toads sickened and died (Shine 1991c). Thus we can expect the imminent presence of toads to exert at least mild selection on keelbacks.

Heritability

Full-sib heritability estimates also include portions of dominance and environmental variance and are thus likely to overestimate heritability

(Falconer and Mackay 1996). I have no information on the contribution of dominance variance and am forced to make the common assumption that it is small. I was able to remove variance attributable to maternal mass however, and incubation treatments appeared to have little influence on toxin resistance.

This result suggests that covariance due to common environment will be minimal in my estimate.

Additionally, it is likely that many of my full-sib groups are, in fact, half- sib groups. Multiple paternity is common in snakes; multiple matings are often

128 observed (including in T. mairii) and molecular data often reveal multiple paternity (e.g. Barry et al. 1992; McCracken et al. 1999). The effect of these cryptic half-sib groups is to give an under-estimate of heritability under a full- sib analysis (Brodie and Garland 1993), making my estimate conservative.

Despite these qualifications, my estimate of heritability for resistance to toad toxin (0.456 ± 0.1) suggests relatively high levels of heritable variation for this trait in this population. Heritability estimates for physiological traits are generally around 0.3 (Roff 1997). Lack of a heritable basis to variation is thus unlikely to be a major impediment to adaptive change in this trait.

Locomotor trade-offs

Locomotor trade-offs were not detected for response to toad toxin, but I found a strong trade-off between locomotor performance and resistance to L. dahlii toxin. This strong trade-off in locomotor performance for L. dahlii toxin contrasts with the minimal or non-existent trade-off detected for toad toxin.

The Humpty Doo keelback population was naïve to toads at the time of testing but has been exposed to L. dahlii for many generations. It is tempting to speculate that selection for increased resistance to L. dahlii toxin has resulted in resistance levels being driven upwards by selection to the point where they are balanced by trade-offs. The lack of trade-off for toad resistance may be due simply to the fact that resistance to toads is not yet under directional selection in this population.

129

Toxin trade-offs

The power analysis suggests that the data have excellent power to detect reasonable levels of correlation between responses to the two toxins. My results thus clearly indicate a very poor correlation (if any) between an individual snakes’s response to toad and L. dahlii toxins. Given that the action of the two toxins is likely to be different this result is not altogether surprising. However, it suggests that selection for resistance to L. dahlii toxin will not pre-adapt a population for resistance to toad toxin. The lack of correlation also suggests that there is likely to be no trade-off between resistance to these two toxins.

That is, selection for increased resistance to one toxin does not equate to reduced ability to deal with the other toxin, at least in keelbacks. In this system at least, snakes appear free to evolve resistance to toad toxin without sacrificing resistance to the toxins of native prey.

While pre-adaptation in terms of resistance to toad toxin is unlikely, sympatry with an extremely toxic prey item may cause selection for other traits that will pre-adapt a population to the invasion of the toad. Most important of these is a change in prey preference or foraging tactics – the presence of L. dahlii may have lead to the evolution of “fussiness” in attack and feeding responses.

Such evolved predator tactics may reduce the impact of the toad or allow a more rapid recovery following the toad invasion. Further research is required to assess these possibilities.

The potential for adaptation

130

Humpty Doo keelbacks exhibit significant and relatively high heritability for resistance to toad toxin. Thus, in the absence of trade-offs or constraints, toxin resistance is likely to respond to selection. I found no evidence for either a trade-off or pre-adaptation due to the presence of the native frog, Litoria dahlii.

Thus, snakes sympatric with L. dahlii are unlikely to have either an advantage or disadvantage in terms of an adaptive increase in toxin resistance.

Additionally, in keelbacks there appears to be no intrinsic trade-off between locomotor performance and resistance to toad toxin suggesting no impediment to adaptive change through this factor either.

Thus I find evidence for heritable variation, probable mild selection and no obvious trade-offs for resistance to toad toxin in keelbacks. The clear inference is that this species is capable of adaptively responding to toad invasion by increasing toxin resistance. The relatively short generation time of keelbacks (<18 mo: Brown and Shine 2002) suggests that rapid adaptive response will be possible in this species. In keeping with this prediction, keelbacks (but not most other frog-eating snake species) remain abundant in areas that toads have occupied for > 50 years.

While rapid adaptive response is possible for toxin resistance, it remains possible that one or more of the other three traits listed (Introduction) may be more labile and respond more rapidly to the presence of toads. The current results simply mean that an adaptive response is possible in keelbacks. To persist through environmental change, one adaptive solution is all that is necessary.

131

My approach, using experimental data, is a level of abstraction away from direct measurement of population-level impact but clarifies the likelihood of an adaptive response. Also, by examining factors that may be influential

(e.g. the presence of a native, toxic frog) we can make a broader level of inference, beyond the particular study species. For example the lack of relationship between resistance to the two toxins in the keelback suggests that a similar lack of relationship is likely in other species (i.e. the toxins are very different). Thus the presence of the toxic native is likely to have little bearing on resistance to toad toxin for other snake species in the community.

Examining the problem of environmental change in this manner places individual studies within the broader unifying framework of evolutionary theory. Ultimately, this should allow meaningful comparisons to be made across species and systems. Such a unified approach is necessary for furthered understanding of the impacts of invasive species and the consequences of environmental change generally (Caughley 1994; D'Antonio and Kark 2002).

ACKNOWLEDGEMENTS

I am grateful to the staff at Beatrice Hills farm for their generous hospitality.

Greg Brown collected morphological data for mothers and set up the incubation treatments. Richard Shine, Greg Brown, Stuart Pimm and two anonymous referees provided comments on an earlier draft. I thank Megan Higgie and

Mark Blows for advice on REML.

132

133

CHAPTER 6

AN INVASIVE SPECIES CAUSES RAPID

ADAPTIVE CHANGE IN AN IMPACTED

NATIVE SPECIES: BLACKSNAKES AND CANE

TOADS IN AUSTRALIA

134

ABSTRACT

Rapid environmental change as a consequence of human activities has led to a heightened extinction rate. It is increasingly apparent however, that some species are able to adapt to these anthropogenic modifications. Additionally, the pace of change varies with the specific impact considered, such that some categories of environmental change may be more likely to elicit an evolved response in impacted populations.

Invasive species represent an instantaneous environmental change (i.e. they are either present or absent). Understanding the potential for species to adapt to such a modification potentially sheds light on the likelihood of species adapting to more gradual changes (such as global climate warming). Here I examine Australian blacksnakes

(Pseudechis porphyriacus) to determine whether they have mounted an adaptive response to the invasion of a lethally toxic prey item, the cane toad (Bufo marinus).

Snakes from toad-exposed localities showed increased resistance to toad toxin and a decreased preference for toads as prey. In separate laboratory experiments I was unable to teach naïve snakes to avoid toxic prey, nor was I able to increase snake resistance to toad toxin through repeated sub-lethal doses. These observations and experiments strongly suggest that black snakes have exhibited an evolved response to the presence of toads. As toads only arrived in Australia in 1935, these evolved responses are rapid, occurring in less than 23 snake generations.

135

INTRODUCTION

The current biodiversity crisis and high species extinction rate is a consequence of rapid environmental change, mediated by human activities

(Ehrlich 1995). Overharvesting, invasive species and altered climate are all examples of significant environmental change. Because such modifications are usually directional, we can expect non-random impact on species affected by a specific change. Non-random impact, of course, equates to Darwinian selection. Therefore, many of the environmental changes mediated by humans exert strong selection on affected species. It is increasingly apparent that evolutionary responses to strong selection can occur rapidly, on time scales traditionally thought of as “ecological” (Hendry and Kinnison 1999; Stockwell et al. 2003; Thompson 1998).

Understanding which species are likely to adapt to a given change provides valuable information for the setting of conservation priorities.

Additionally, some categories of environmental change may facilitate adaptive responses by impacted species. In short, understanding the potential for evolution to affect the outcomes of anthropogenic impact allows us to further refine conservation priorities and strategies (Ashley et al. 2003).

Invasive species are a major concern to conservationists (Mack et al.

2000). They are second only to climate change in ubiquity and there are several examples of invasive species driving native species to extinction (e.g. Ogutu-

Ohwayo 1999). The arrival of an invasive species represents a much more

136 abrupt environmental change than do more gradual processes, such as global warming. Thus, from an evolutionary perspective, invasive species often represent an instantaneous and strong change in the selective environment. If species can adapt to such instantaneous change, it seems more likely that they can also respond to relatively slow changes (such as climate warming). What is the possibility that native species can adapt to an invader? Obviously this will depend, among other things, upon the strength of the selection pressure and

(given the instantaneous nature of the change) the amount of heritable variation at traits mediating the impact in native species. Here I examine the possibility of an adaptive response by an Australian snake to the invasion of a toxic prey item, the cane toad.

Toads were introduced into Australia in 1935 (Lever 2001). They are highly toxic and the principal toxin is unique to toads (Chen and Kovarikova

1967). As Australia has no native species of bufonids (Cogger 2000; Lutz 1971) the arrival of toads presents a highly toxic potential prey item to a naïve predator fauna. Australian snakes, in particular, have faced massive impacts in the presence of toads. More than 49 species of snake have the potential to be impacted, and almost all of these are poorly equipped to survive a likely dose of toad toxin (Phillips et al. 2003). The arrival of toads thus imposes selection on at least three traits: physiological resistance to toad toxin, prey preference

(the tendency to eat toads) and the morphology of impacted snake species

(relatively small-headed snakes are less likely to be capable of consuming a toxic dose: Phillips and Shine submitted-b).

137

One native species facing a high impact from toads is the red-bellied blacksnake (Pseudechis porphyriacus). This relatively large elapid feeds primarily on frogs and has very low resistance to toad toxin (Phillips et al. 2003).

Anecdotal reports indicate massive declines in black snake populations following the arrival of toads (Covacevich and Archer 1975; Fearn 2003; Phillips and Fitzgerald 2004; Rayward 1974). The current distribution of the black snake includes areas of sympatry and allopatry with toads. Here I compare toad- naïve and toad-exposed populations of black snakes to examine the possibility that these snakes display an adaptive response to the presence of toads.

138

METHODS

Comparing toad-exposed and toad-naïve populations

1) Prey Preference.

Twelve black snakes were collected from each of the following categories: (1) populations exposed to toads for 40-60 years (Childers and

Agnes Waters, Qld) and, (2) toad-naïve snakes from two populations; one immediately adjacent to the expanding toad front (Casino, NSW) and one approximately 300km from the front (Macquarie Marshes, NSW). After a minimum of 3 weeks in captivity, each snake was offered a frog (Limnodynastes peronii – a widespread species, sympatric with snakes at all collection localities) and a toad in random order, three days apart. Prey items were offered to the snakes freshly killed, to eliminate behavioural differences between the prey and so I could remove the parotoid glands from the toads; not doing so would likely have resulted in the death of snakes during the course of the experiment. In each case the snakes were left undisturbed for 24 hours, after which I recorded whether or not the prey item had been eaten. Differences in numbers of prey consumed between toad-exposed and toad-naïve localities were compared using Fisher’s exact test.

2) Toxin resistance.

Fourteen snakes from toad-exposed localities and 24 from toad-naïve localities were tested for resistance to toad toxin. These snakes included fifteen

139 tested previously for prey preference (5 from toad-exposed areas and 10 from toad-naïve areas) and additional animals collected from a range of localities within each category. Toad-exposed snakes came from a range of populations representing exposure times of 5-60 years.

Resistance to toad toxin was assayed in a manner identical to that reported in Phillips et al. (2003), using the same toxin extract. Toad toxin was obtained from skins of freshly killed cane toads collected from the Lismore area

(northern NSW). A single extraction of toad toxin was made for the entire study, to remove among-toad variance in toxicity and to accurately control dosing. The resistance of individual snakes to toad toxin was assayed using the decrement in swimming speed following a dose of toxin (methodology modified from Brodie and Brodie 1990). A large percentage reduction (%redn) in swimming speed indicates a lower resistance to toxin than a smaller reduction in swimming speed.

Each snake was given a dose of 80μg of toad skin per gram of body mass

(a dose previously calculated to be non-lethal but provide measurable reductions in speed: Chapter 1). Dosing was achieved with a feeding tube attached to a syringe or calibrated micropipette, inserted into the snake’s stomach to a depth of 30% of its snout-vent length.

Differences in %redn were compared between exposed and naïve populations using a t-test. Additionally, I examined the relationship between

%redn and time since exposure to toads in toad-exposed populations by simple linear regression.

140

Learning and acquired resistance experiments

Observed differences in resistance and prey preference either could be acquired during a snake’s lifetime, or be the result of changes in gene frequency due to adaptive evolution. Following the observation of differences between exposed and naïve populations, I exposed captive snakes to toad toxins to evaluate the possibility of either a learnt response or acquired resistance in naïve black snakes.

1) Learning.

Sixteen snakes were collected from toad-naïve areas. Snakes were kept in captivity for a month before the learning trial commenced. Each snake was offered two prey types, three times each in random order. I used laboratory mice and a lizard (Eulamprus tympanum, Scincidae; an allopatric and hence novel species to all my snakes) as my two prey types. Prey items were offered to each snake dead and one at a time. Snakes were left undisturbed for 24 hours, after which I recorded whether or not the prey item had been eaten.

Successive feedings were four days apart. Following these six feeding events, a prey item was introduced that contained a high but sublethal dose of toad toxin

(120μg of toad skin per g of body mass, 65% of LD50; a dose that reduces a snake’s locomotor ability by more than 50% for >24 hours). Eight snakes received a toxin-laced lizard and eight snakes received a toxin-laced mouse.

Following the consumption of this prey item, I repeated the previous feeding

141 schedule, recording the number of prey consumed. At the conclusion of the trials I counted the number of prey items of each type that were consumed before and after the dose of toxin and used repeated measure ANOVA to assess differences in these scores.

142

2) Acquired resistance.

Twenty snakes from toad-naïve populations (including those used in the learning trials) were assessed for the possibility of acquired resistance. Every five days, I administered either a dose of toxin or a dose of water to each snake.

This was repeated four times so that half the snakes received four doses of toxin and half received four doses of water. Dosing method was identical to that described for assessment of differences in resistance to toxin between exposed and naïve populations (dosing rate: 60μg per gram of body weight, dose = 32% of the LD50). Four weeks after the last dose, all snakes were assessed for resistance to toxin in a manner identical to that used to compare exposed and naïve populations. Following the calculation of %redn scores, I compared the toxin and control groups and also compared all snakes in this experiment with snakes assessed for resistance from naïve localities. This allowed me to test whether repeated doses of toxin increase resistance and also whether a single dose of toxin (sixteen snakes had received at least one dose of toxin in the learning experiment) could increase resistance.

143

RESULTS

Comparing toad-exposed and toad-naïve populations

1) Prey preference.

All the snakes, from each exposure category, ate the frog that was offered to them (Figure 1). Exactly half (i.e. six) of the snakes from toad-naïve populations consumed a toad, and no toads were consumed by snakes from toad-exposed populations – a significant difference (χ2 = 5.6, df = 1, p = 0.014;

Figure 1).

2) Toxin resistance.

A significant difference between exposed and naïve populations was detected with naïve populations exhibiting higher %redn and thus, lower resistance to toad toxin (exposed mean = 16%, naïve mean = 32%; two-tailed t- test, unequal variances; t = 3.112, df = 24, p = 0.005, Figure 2). Additionally,

%redn scores decreased (and hence level of resistance increased) with exposure time (F1,12 = 7.72, p = 0.017, Figure 3). Importantly, the y-intercept of the regression of %redn on exposure time yielded a value of 34%, a very similar value to the mean %redn of naïve populations (32%).

144

100

Frogs eaten (%) 75 Toads eaten (%)

50

25 Percentage eaten

0 yes no

Exposure to toads

Figure 1. The percentage of snakes from toad-exposed and toad-naïve populations willing to eat a toad or a frog. Twelve snakes from each exposure category were used. No snake from a toad-exposed locality would consume a toad. Error bar represents one standard error.

145

100

75

50

25

0 Reduction in speed (%)

-25 Naive Exposed

Exposure category

Figure 2. Resistance to toad toxin in toad-exposed and toad-naïve populations. Open circles represent individual snakes, cross bars represent the mean for each category. A large percentage reduction in speed indicates low resistance to toxin. Hence snakes from toad-exposed populations exhibited higher resistance to toad toxin.

60

40

20

0 Reduction in speed (%)

-20 0 10 20 30 40 50 60 70

Exposure time (years)

Figure 3. Resistance to toad toxin as a function of the time a population has been exposed to toads. A large percentage represents a low resistance. Hence resistance to toad toxin increases with increasing exposure time.

146

Learning and acquired resistance experiments

1) Learning.

To compare numbers of prey eaten before and after the administration of toxin required a repeated measures ANOVA with prey type and toxin/control

(treatment) as orthogonal factors and number of prey eaten (before and after administration of the treatment) as the dependent variables.

This analysis revealed that snakes fed a toxic prey item showed no inclination to avoid the prey item in further feeding opportunities (F1,27 = 0.007, p = 0.93,

Figure 4). This effect was independent of prey type (interaction, F1,26 = 0.1678, p

= 0.69) and suggests that a black snake surviving an encounter with a toad is unlikely to avoid toads in the future.

2) Acquired resistance.

After four doses of toxin over the period of a month there was no change in the level of resistance exhibited by snakes when compared with a control group (F1,18 = 2.95, p = 0.10) and mean %redn in swimming speed was higher in the toxin-exposed group (42%), than the control group (22%); that is, snakes given several doses of toad toxin tended to exhibit lower rather than higher resistance to toxin. Because most of the snakes involved in this experiment had previously been exposed to a single dose of toxin during the learning experiment, it is possible that this single dose changed their resistance. To assess this possibility I compared the resistance of all the snakes in this experiment (n=20, mean %redn = 32.16) to the resistance of all snakes

147 previously tested from naïve populations (n=24, mean %redn = 32.42).

ANOVA revealed no significant difference in %redn between these two samples either (F1,42 = 0.0016, p = 0.97).

4 Control

Toxin 3

2

1 Mean number of prey taken

0 Before After

Figure 4. Can snakes learn to avoid toxic prey? The number of prey taken by snakes before and after exposure to a toxic prey item. Two prey types were used, only one of which was laced with toxin for each snake (see methods).

148

DISCUSSION

My results show differences between blacksnakes from toad-exposed versus toad-naïve populations in their physiological resistance to toad toxin and their willingness to eat toads. Importantly, both of these differences are in an adaptive direction; that is, we see an increased resistance to toxin and lowered preference for consuming toads in toad-exposed populations. These changes either could be plastic changes (acquired within an individual snake’s lifetime) or evolved changes (a genetically-coded response to the strong selection imposed by toads). To discriminate between these possibilities, I attempted to elicit acquired responses in toad-naïve captive snakes. However, I found no evidence that snakes can learn to avoid a toxic prey item, nor that they can acquire physiological resistance to toad toxin. My inability to elicit acquired responses in either of these two traits suggests that the differences observed between toad-exposed and toad-naïve populations are due to adaptation rather than phenotypic plasticity.

The interpretation of an evolved response depends on the degree to which my acquired resistance and learning experiments mimic reality. In designing these experiments I operated under the premise that most snakes will be lucky to survive an encounter with a large toad (Phillips et al. 2003, Seabrook and Fitzgerald unpub. data, Shine unpub. data). That is, a black snake that eats a large toad is likely to die. The window of sub-lethal toxin effect is relatively narrow (Phillips et al. 2003), such that few toads will be large enough to cause

149 illness but small enough to be non-lethal. Because it is unlikely, therefore, that an individual snake will have several chances to learn avoidance, my learning experiment was based on a single noxious encounter. The fact that black snakes did not learn to avoid toxic prey is surprising given that Burghardt et al. (1973) and Terrick et al. (1995) elicited learnt aversion in gartersnakes (Thamnophis spp.) after a single toxic encounter. Gartersnakes are often sympatric with toxic amphibians and may have evolved this learning capacity as a response to toxic prey (Brodie and Brodie 1999a). In contrast, blacksnakes are not known to be sympatric with any naturally-occurring, dangerously toxic prey and thus may have been under little or no selection to learn avoidance. It remains possible, however, that some cue specific to toads could increase a snake’s tendency to learn avoidance (much as aposematic colouration appears to enhance learned avoidance in gartersnakes (Terrick et al. 1995)). If this is the case, learning could occur but my experiment wouldn’t elicit it. Given that Australian snakes have no evolutionary history with toads, or their toxins, and toads are not aposematically coloured, it seems unlikely (but not impossible) that a toad- specific cue would increase learning ability. Nevertheless, if we assume that naïve blacksnakes are unable to learn avoidance despite a near-lethal encounter, the observation of strong differences in prey preference between toad-naïve and toad-exposed populations implies an evolved response. Whether this response is a congenital disposition to avoid toads or an evolved ability to learn from a single noxious encounter remains to be seen. Prey-preference has a highly heritable basis in gartersnakes (Arnold 1981; Arnold 1992) and so it seems likely

150 that this trait will also have high heritability in blacksnakes. Additionally, it is important to note that there is variation for the tendency to eat toads in naïve populations of blacksnakes (only half of the naïve snakes consumed a toad, Fig.

1). The heritability of learning ability, however, has never been measured and I detected little variation in this trait in naïve populations of black snakes

(although sample sizes were relatively small). Further work exploring the basis of the change in prey preference would be enlightening.

While most snakes probably only get a single chance with a large toad, it is possible that they could consume several small toads at different times with minimal ill effect, and acquire an increased level of resistance through an immune or other physiological response. The acquired resistance experiment thus exposed snakes to four sub-lethal doses. One month after these dosings, toxin-exposed snakes were no better equipped to deal with toad toxin. Again, this result suggests that the differences between exposed and naïve populations are probably evolved rather than acquired.

In light of the prey-preference results (no snake from toad-exposed areas consumed a toad), it superficially seems paradoxical that I also detected evidence of selection on toxin resistance. This difference may be the result of historically strong selection when toads first arrived and toad-avoidance had yet to become fixed (or nearly fixed) in the population. Alternatively, if the prey preference result reflects an evolved ability to learn avoidance of toxic prey, there may be ongoing selection on toxin resistance. Additionally, spatial and temporal variation in relative prey abundances and/or levels of snake

151 preadaptation, would also lead to concurrent evolution in resistance and prey preference (Brodie and Brodie 1999a; Gomulkiewicz et al. 2000).

Recent work also documents a reduction in the relative head size of blacksnakes (and hence their ability to eat large prey items) as a consequence of exposure to toads (Chapter 4). Thus it appears that blacksnakes show adaptive change in multiple traits in response to the presence of toads. Given that the generation time of blacksnakes is approximately 3 years (Shine 1978) and that toads have only been present in Australia for <70 years, these adaptive changes are rapid (occurring in < 23 generations).

The current study is one of the first to demonstrate adaptation by a native species in response to an impact of conservation concern from an invasive. As such it places invasive species into a growing list of environmental changes to which adaptive response has been demonstrated (Stockwell et al.

2003). That some species at least are capable of mounting such a rapid adaptive response to an instantaneous change in the environment suggests that other, more gradual changes (such as global warming) may also elicit adaptive responses rather than extinction. This result highlights the importance of considering the potential for adaptation when predicting the long-term impact of environmental change and also highlights the need to maximise the adaptive potential of managed species through the maintenance of large, genetically diverse populations.

ACKNOWLEDGEMENTS

152

This study would not have been possible without the assistance of many people who helped with the collection of snakes (a difficult undertaking in toad-exposed areas). Foremost among these is Ian Jenkins with additional help from Eric Bateman, Julie Dickson, David Fouche, Richard Ghamroui, Jeff

Hayter, Andrew Hugall, Ray Jones, Michael Kearney, Amanda Lane, Clare

Morrison, Adnan Moussalli, Luke Shoo, Devi Stuart-Fox, Eric Vanderduys and

Michael Wall. Steve Phillips and Jai Thomas assisted with husbandry and maintenance. Richard Shine and Michael Wall reviewed and improved an earlier version of this chapter. Funding was provided by grants from the

Australian Research Council, The Royal Zoological Society of NSW, The Royal

Linnean Society of NSW and the Norman Wettenhall Foundation.

153

CHAPTER 7

THE MORPHOLOGY, AND HENCE IMPACT, OF

AN INVASIVE SPECIES (THE CANE TOAD,

BUFO MARINUS) CHANGES WITH TIME SINCE

COLONISATION.

154

ABSTRACT

It is increasingly apparent that the success of a particular invasive species may be related to its phenotypic lability. Despite often strong founder effects and concomitant reductions in genetic diversity, many successful invasive species still exhibit adaptive change in response to their new environment. Successful invaders are also a major global conservation problem. To understand the likely long-term impacts of a particular invader it is critical that long-term changes in the invader’s phenotype are assessed, as these changes may well influence the level of impact the invader has on native species. Here I examine morphological change, as a consequence of time since colonisation and other spatial factors, in the cane toad (Bufo marinus). Cane toads are highly toxic and have had a major impact on Australian native predators since they were introduced in 1935; naïve predators die attempting to consume them. The amount of toxin that a predator is exposed to depends upon both the body size of the toad and also the relative toxicity of the toad (here measured by the relative size of the toad’s parotoid glands). Using multiple regression and a model-selection approach, I discovered that both toad size and relative toxicity decrease with time since colonisation.

This shows first, that toads, like many other successful invasives, exhibit high phenotypic lability. Additionally, this result strongly suggests that the impact from toads on predators decreases as a consequence of time since colonisation.

155

INTRODUCTION

Invasive species are increasingly being used as model systems within which to study adaptation and plasticity (Lee 2002). Paradoxically, many invasive species appear to exhibit adaptive responses to new environments despite what superficially appear to be strong founder effects and concomitant reductions in genetic diversity. Given that successful invaders represent a biased sample of all potential invaders, it seems increasingly likely that the adaptive potential of invaders is strongly linked to their probability of success in a new environment (Kolbe et al. 2004; Parker et al. 2003; Simons 2003;

Stockwell et al. 2003; Thompson 1998). Thus, understanding the evolutionary processes underlying successful invasions is of both broad theoretical and practical interest.

Invasive species also often have large impacts on native communities and, given their ubiquity, are regarded as a major threat to global biodiversity

(Mack et al. 2000; Williamson 1996). Thus, examining change in invasive species may also give us insights into the long-term impact of a particular invader: by examining change in traits that mediate an invader’s impact we can understand how the level of impact changes through time.

Cane toads (Bufo marinus) are extremely successful invaders throughout the Caribbean and Pacific, having successfully invaded more than 20 countries to date (Lever 2001). Their colonisation history from their native range in South

America is very well documented (Easteal 1981): by the time toads were

156 brought to Australia in 1935, they had already undergone several founder events and genetic diversity as measured at allozyme and microsatellite markers was much reduced (Estoup et al. 2001). They are a large, toxic anuran, and since their initial release they have spread to occupy more than one million square kilometres of the Australian continent (Lever 2001). Among other potential impacts, toads are known to massively impact native terrestrial predators, which are naïve to toad toxin and die attempting to ingest them

(Burnett 1997; Covacevich and Archer 1975; Oakwood 2003; Phillips et al. 2003;

Smith and Phillips submitted). The dose of toxin that a predator will be exposed to in an interaction (and hence the risk of death to the predator) depends upon two factors: the body size and the relative toxicity of the toad.

Large toads contain greater quantities toxin than small toads and even at the same body size, some individuals will be more toxic than others (Chapter 3).

While size can be measured directly, relative toxicity may be more complicated.

Conveniently, most toxin in the skin of toads is stored in the large parotoid glands located above the shoulders (Meyer and Linde 1971). Thus, we can use the size of the parotoids as an index of the amount of toxin carried by a toad.

Here I examine changes in these two aspects of toad morphology, body size and relative parotoid size, as a consequence of time since colonisation.

157

METHODS

Collection of morphological data

I measured all of the 140 cane toads present in the collection of the

Queensland Museum. This specimen series represented animals collected since

1935 (the year of toad arrival). Each individual was measured for snout- ischium length (SIL), parotoid gland length (PL) and parotoid gland width

(PW). Information on collection locality and date of collection was also taken from the museum database.

Collection of data on time since colonisation

More than 2000 records of toad locality and date were available from the

Queensland Museum and from the dataset collected by Floyd et al. (1981).

Sabath et al. (1981) and Easteal et al. (1985) used the latter dataset to map the spread of toads in Australia, however the results were hand drawn maps of the toad distribution at five yearly intervals. Improvements in mapping tools since then (i.e. GIS) allowed me to create a single digital map, of far greater accuracy, which can be used to provide information on the toad expansion at yearly intervals. To do this I used linear interpolation of locality dates in ARCVIEW to derive a layer describing the arrival date of toads; details of the process can be found in Chapter 4.

Following the derivation of this surface, the Queensland Museum toad locality records were plotted and the year of toad arrival at each site was

158 extracted. For each measured toad I subtracted the year of toad arrival (from the GIS layer) from collection year (from the Queensland Museum database) to yield time since colonisation (TSC) – that is, the number of years a population of toads had been present in an area at the time a toad was collected.

Collection of climatic data

Because my data has a spatial component, I attempted to account for as many spatially varying factors as possible to reduce the potential for a spurious correlation and to reduce error variances. In addition to Latitude (DecLat, from the Queensland Museum database), I derived several climatic layers for

Australia using the program ANUCLIM (Hutchinson et al. 1999) and a digital elevation model of Australia with 0.05º grid cells. Toad locality data were laid over the resultant climate grids in ARCVIEW and I extracted the climatic data for each locality using the ARCVIEW extension BIOCLIMav (Moussalli 2003). I used several climatic variables that are likely to influence toad morphology: annual mean temperature (AMT), minimum temperature of the coldest period

(MinTempCP), annual precipitation (APrecip), precipitation seasonality

(PrecipSeas), moisture index seasonality (MoisIndSeas) and annual mean humidity at 3pm (AMHumid).

Data analysis

I examined the effect of TSC on toad morphology using a model selection approach. By testing the relative information content of all possible

159 models, I determined whether TSC was an important factor (i.e. was it present in the best model/s) and also selected the most parsimonious model (i.e. the model that explained the most variance with the least number of factors) describing toad morphology.

Because many of the climatic variables were correlated to varying degrees, and to reduce the number of factors, I calculated the first three principal components of climatic and latitude variables. Two analyses were run: The first used toad snout-ischium length (SIL) as the dependent variable and the second used toad parotoid size. Parotoid size (PS) was calculated as the first principal component of the two parotoid size variables I measured (PL and

PW). The multiple regression for toad parotoid size also included toad body size (SIL) as a fixed independent variable as I was only interested in changes in relative parotoid size. I log-transformed all variables prior to the calculation of principal components and the TSC variable was mean-centred ( y'= y − y ) prior to analysis. Mean-centring (such that the new mean is zero) ensures that estimated coefficients are informative even in the presence of interactions; this method also reduces colinearity between variables and their interaction terms

(Jaccard and Turrisi 2003). Mean centring was not necessary for principal components because their mean was already zero.

With four independent, non-fixed variables I had 15 combinations of primary variables that could produce a model (ignoring interaction terms). To make model exploration and interpretation tractable I only examined first order interactions between factors. Each of the 15 combinations was run as a full

160 model and I deleted interaction terms if p-values indicated they were not significant (i.e., p > 0.05). For each combination of primary variables I thus derived the most parsimonious reduced model and I calculated the Akaike information criterion (AIC) value for this model. I collected the best set of models for each species and each independent variable based upon these AIC values, with models <2 units from the best model (i.e. Δi < 2) retained within the best set (Burnham and Anderson 2001). All statistical analyses were performed in JMP (v5).

Some of the models thus selected contained interaction terms. My primary interest was whether time had an important influence on toad morphology and, if so, the direction of the effect. The presence of interaction terms complicates the interpretation of main effects because the partial coefficient for the main effect of interest depends on the values of other variables. Mean-centring causes the main effect coefficient to be calculated for the mean value of interacting variables. However, in all models with interaction terms affecting the coefficient of TSC, I also calculated a range of coefficients using values for the interacting variables that were two standard deviations above and below their mean.

161

162

RESULTS

Calculation of multivariate model components

The first three principal components of climatic and latitude variables accounted for almost 95% of the variation in these seven factors (Table 1).

Eigenvectors indicate that PCClimLat1 is principally a latitude/temperature axis and captures most of the latitudinal variation in the environmental factors.

PCClimLat2 appears to be principally a precipitation axis, capturing the resultant variation in humidity and moisture index seasonality. PCClimLat3 is more difficult to interpret but may be capturing altitudinal variation in the environmental variables.

For toad parotoid size (PS), the first principal component of PW and PL captured more than 99% of the variation in both these variables (reflecting their strong correlation) with equal loadings on both.

Toad snout-ischium length

The single best model describing toad SIL included all but one of the independent variables and accounted for more than 20% of the variation in toad

SIL. Time since colonisation (TSC) has a negative effect on toad SIL in this model, however its effect is modified by an interaction with PCClimLat2 – the precipitation axis: At high values of PCClimLat2 (approx. two standard deviations above the mean) the negative effect of TSC is reduced or reversed.

163

Functionally, this means that toad body size decreases less rapidly after colonisation in wetter (higher precipitation) areas than in drier areas.

Table 1: Results of principal component analysis of climate and latitude variables for cane toad records from the Queensland Museum.

Principal components: PCClimLat1 PCClimLat2 PCClimLat3 Eigenvalue 3.96 2.40 0.27 Cumulative percent 56.57 90.88 94.68

Eigenvectors Ln(AMT) 0.463 -0.082 0.610 Ln(Aprecip) 0.130 0.612 -0.097 Ln(PrecipSeas) 0.471 -0.147 -0.393 Ln(MoisIndexSeas) 0.359 -0.400 -0.195 Ln(MinTempCP) 0.433 0.254 0.457 Ln(AMHumid) 0.088 0.610 -0.222 Ln(DecLat) -0.471 -0.017 0.410

Abbreviations are as follows: AMT, Annual mean temperature; Aprecip, Annual precipitation; PrecipSeas, Precipitation seasonality; MoisIndSeas, Moisture index seasonality; MinTempCP, Minimum temperature of the coldest period; AMHumid, Annual mean humidity at 3pm; DecLat, Decimal latitude. Eigenvector weights > 0.3 are shown in bold.

Table 2: Parameter estimates for the best models describing toad body size (SIL) and relative parotoid size as a function of climate and latitude (PCClimLat1, 2 and 3) and time since colonisation (TSC).

Factor Snout-ischium length Parotoid size

Intercept -0.0157 -0.0011

Snout-ischium length - 2.5853 PCClimLat1 -0.0491 0.0275 PCClimLat2 0.0874 0.0220 PCClimLat3 - -0.1182 Time since colonisation -1.5584 -0.4617

PC1 x PC2 - -0.0314 PC2 x TSC 0.5273 -

r2 0.204 0.985 TSC range -3.18 - 0.07 -

164

Coefficients significantly different from zero are shown in bold. For brevity, PCClimLat has been further abbreviated to PC for interaction terms and time since colonisation has been abbreviated to TSC. TSC range is the partial coefficient calculated for TSC based on values of interacting variables two standard deviations from the mean.

Toad parotoid size

The single best model describing variation in relative parotoid size was also complex, involving all independent variables (Table 2). In this model time since colonisation had a negative effect on relative parotoid size, unmodified by interactions.

165

DISCUSSION

My results show a significant effect of climate, latitude and time since colonisation on toad morphology. Interestingly, both snout-ischium length

(SIL) and relative parotoid size models indicate that time since colonisation

(TSC) has a significant effect on toad morphology, independent of other spatially varying factors. Time since colonisation appears to be associated with a reduction in both overall size of toads and in the relative size of their parotoid glands. That toad morphology should be so strongly influenced by recent colonisation history is an important result and a reminder of the importance of recent history in invasion events.

For toad body size, my model contained an interaction that modified the magnitude (and eventual direction) of the effect of TSC. In extremely high rainfall areas (around two standard deviations above average), the negative effect of TSC on toad size is reduced or reversed.

Why do large, big-glanded individuals become less common through time? There are two possible reasons why this may be so and they are not mutually exclusive. First, toads may change their environment through time

(e.g. by depleting food resources) such that attaining large size and maintaining costly structures (which poison glands presumably are) becomes increasingly difficult. Freeland (1986) found that toad densities and body condition were lower in long-established populations compared with newly established populations. Interestingly however, the change in body condition was not

166 associated with a difference in the size of fat bodies, indicating that lack of food was not necessarily driving the change. Unfortunately, Freeland’s (1986) study was conducted over a single year, so it is possible that local variation in conditions in that year contributed to the patterns he observed. Certainly, studies by Alford and colleagues (1995) in the same areas a few years later reported the opposite pattern in density.

The second possible explanation for changes in toad morphology with

TSC is that toads may exhibit adaptive change to a new environment (e.g. reduced predation pressure, following the extirpation of predator populations) where large body sizes and poison glands confer little or no selective advantage. There are certainly many instances documenting apparent adaptation by invaders to a novel environment (e.g. Grosholz and Ruiz 2003;

Losos et al. 1997; Simberloff et al. 2000). While my data are unable to unequivocally support either scenario, they do show that Australian toads are morphologically labile. This lability may be an important factor in their success as an invader. In fact it may be that most successful invaders are successful partly because they exhibit phenotypic lability in response to new environments (Lee 2002; Parker et al. 2003).

Irrespective of the exact mechanism causing morphological change in toads, either scenario assumes a cost associated with producing a large body or a large parotoid gland. Areas of high precipitation are undoubtedly favourable environments for cane toads (given that their native range encompasses the

Amazon Basin) and it seems logical that physiological costs associated with

167 developing a large body or large glands will be minimised in favourable environments. This reasoning probably explains the presence of an interaction between TSC and precipitation in my SIL model.

My data also suggest that toad body size and relative parotoid size exhibit latitudinal clines (PCClimLat1 is primarily a latitude/temperature axis).

Interestingly, the models for body size indicate an increase in toad body size with increasing latitude/decreasing temperature, in line with Bergmann’s rule for endotherms and against the general pattern in body size clines for ectotherms (although only insects, squamate reptiles and turtles have been examined in any detail, Ashton and Feldman 2003; Mousseau 1997).

Additionally, there is no support for an interaction between time since colonisation and latitude/temperature. If the cline in toad body sizes was an evolved effect, as has been demonstrated for latitudinal clines in Drosophila body size (Gilchrist et al. 2004; Huey et al. 2000), we would expect an interaction between TSC and PCClimLat1. The absence of this interaction suggests that the cline in toad body sizes shown here is a consequence of developmental plasticity rather than evolution, as the cline is present irrespective of the length of time since colonisation.

Because the quantity of toxin carried by a toad is a factor of the toad’s body size and relative parotoid size, reductions in these traits will translate into a reduced impact on predators. Thus, the change in morphology associated with TSC indicates that, except in the wettest areas of the toad’s current distribution, the level of impact imposed by toads on predators will decrease

168 with time since colonisation. Whether this level of impact remains lowered will depend upon the exact mechanism driving the morphological change.

Nevertheless, my results show the importance of considering the possibility of rapid phenotypic change in an invader when assessing the long-term impact of the invader on a native community. It is increasingly becoming apparent that successful invaders are successful partly because they exhibit phenotypic lability in response to a new environment. If this is true, then examining rapid phenotypic change at traits influencing an invader’s impact on natives is an important prerequisite to understanding the long-term impact of an invader.

ACKNOWLEDGEMENTS

I thank Patrick Couper, Andrew Amey and Heather Janetski at the

Queensland museum for access to specimens, entertaining discussion and cups of tea. Richard Shine provided useful discussion and comments on an earlier draft. Funding was provided by the Australian Research Council and the

Norman Wettenhall Foundation.

169

CHAPTER 8

SPATIAL AND TEMPORAL VARIATION IN THE

MORPHOLOGY (AND THUS, PREDICTED

IMPACT) OF THE INVASIVE CANE TOAD

(BUFO MARINUS) IN AUSTRALIA

170

ABSTRACT

The impact of an invasive species is unlikely to be uniform in space or time, due to variation in key traits of the invader (e.g. morphology, physiology, behaviour) as well as in resilience of the local ecosystem. The low genetic diversity typical of invasive species suggests that much of the variation in an invading taxon (and thus, in its ecological impact) is likely to be generated by the environment and recent colonisation history. I use a model-selection approach to describe effects of the environment and colonisation history on key morphological traits of an invader (the cane toad, Bufo marinus). These “key traits” (body size and relative toxicity) mediate the impact of toads on Australian native predators, which often die as a consequence of ingesting a fatal dose of toad toxin. Measurements of museum specimens collected over > 60 years and across the state of Queensland show that seasonal variation in toad body size (in turn, due to seasonal recruitment) effectively swamps much of the spatial variance in this trait. However, relative toxicity of toads (measured by size of the poison glands relative to body size) showed strong spatial variation and little seasonal variation.

Thus, the risk to a native predator ingesting a toad will vary on both spatial and temporal scales. For native predators capable of eating a wide range of toad sizes (e.g., quolls, varanid lizards), seasonal variation in overall toad size is likely to be the most significant predictor of risk (and thus, ecological impact of toads). In contrast, gape- limited predators restricted to a specific range of toad sizes (such as snakes) will be most strongly affected by the relative toxicity of toads. Gape-limited predators will thus experience strong spatial variation in risk from toad consumption.

171

INTRODUCTION

Invasive species are a major threat to global biodiversity (IUCN 2001;

Mack et al. 2000). Human-assisted transport and environmental change has broken down biogeographical barriers in many parts of the world leading to species invasions and irrevocable changes to native communities. Although some invaders have little impact on natural ecosystems, other taxa exert varying levels of impact, and this process can lead to local or global extinction

(Ogutu-Ohwayo 1999; Williamson 1996).

Clearly, the magnitude of impact of an invasive species will be different not only for different native species, but also may vary among populations of any given species. Even if we restrict analysis to interactions between a single predator-prey species-pair (for example), the magnitude of impact will be far from constant through space and time. For example, some populations of native taxa may be more or less vulnerable to the threat posed by the invader; and similarly, populations of the invader may differ in traits (such as body size or toxicity) that determine the intensity of their effect on the native system.

Such spatial and temporal variance in traits of the invader may be generated either by plasticity or local adaptation; and if invading populations vary in such ways, this variation needs to be considered by ecologists and conservation biologists attempting to understand or manage the system (Parker et al. 2003).

The combined impact of founder effects and drift at initial small population sizes is expected to leave invaders with low levels of genetic

172 diversity (Allendorf and Lundquist 2003). Although this attribute (among others) has encouraged researchers to utilise invasive species as model systems in evolution (Lee 2002), it also suggests that much of the variation within a population of invaders is likely to be of environmental (rather than genetic) origin. Hence, much of the potential variance in impact from an invader is also likely to be generated by the local environment. Understanding how the environment affects an invader and how this flows on to affect impacted native species has the potential to clarify much of the spatial and temporal variation in the impact of invasive species.

For any invading taxon, the effect of the environment on key traits may depend not only on the environment itself, but also on how long the invader has been present in an area (Chapter 7). A longer duration of time in a given area may allow the invading population to adapt; and also, may allow time for the invader to modify the environment (e.g., by depressing resource levels).

Thus, any analysis of the effect of the environment on an invasive species should incorporate information on the duration of time for which the invader has been present. Factors such as distributional range (Elton 1958), resource availability and coadaptation with native communities (Thompson 1998) may all be strongly influenced by recent history in invasive species.

The above ideas suggest the following approach to assessing the spatial and temporal variation in impact by an invading taxon:

1. First, we need to understand the mechanism of impact, in order to

identify what traits of the invader (and possibly the native species)

173

mediate that impact. These traits may be behavioural, morphological,

and/or physiological. The important challenge is to identify traits for

which changes are likely to modify the level of impact.

2. These traits need to be sampled at different locations and at different

times. At each location, environmental variables that may influence the

trait (including colonisation history) should also be sampled.

3. The effects of the environment and time on each trait should be modelled

so that the resulting model can be mapped onto the area of interest.

Below I describe a case study examining temporal and spatial variation in predicted impact from an invasive species as a consequence of environmental factors, including colonisation history.

Case study: Australian predators and the invasive cane toad

Cane toads are large, toxic anurans native to Central and South America.

Introduced into Australia in 1935, toads have spread throughout large areas of

Queensland and have recently entered the Northern Territory and New South

Wales. This invasive taxon currently occupies a range of more than one million square kilometres within Australia (Lever 2001). Toads reach extremely high densities in suitable habitat (densities >2000 per hectare have been recorded,

Freeland 1986) and have three types of potential impact on Australian native species (Freeland 1987): 1) predation on small animals, 2) competition for food and/or shelter resources and, 3) because they are extremely toxic and the toxin

174 is novel to Australian predators, toads are likely to kill most native predators that attempt to eat them.

Despite these three possibilities, the ecological impact of toads on the native Australian fauna has been poorly elucidated, mainly due to logistical difficulties and a lack of baseline data for comparison (van Dam et al. 2002).

Nevertheless, there is mounting evidence that the bufonid invasion has severely impacted populations of native predators that attempt to eat toads and die as a consequence of ingesting the toxin. Specifically, evidence is accumulating that toads have had a severe impact on native snakes (Covacevich and Archer 1975;

Fearn 2003; Phillips et al. 2003; Phillips and Fitzgerald 2004; Webb et al. in press), varanid lizards (Burnett 1997; Smith and Phillips submitted) and quolls

(a medium sized marsupial carnivore, Burnett 1997; Oakwood 2003; van Dam et al. 2002) all of which die attempting to eat toads. It is this mode of impact that I deal with in the current analysis.

For a predator consuming a toad, the dose of toxin will depend upon two factors: the body size of the toad (bigger toads carry more toxin, Phillips and Shine submitted-b), and the relative toxicity of the toad (even at the same size, some toads will be more toxic than others: Chapter 3). These factors will be of different importance to different kinds of predator taxa. Some species

(e.g. quolls and large varanids) are capable of capturing and consuming toads of a very wide range of body sizes. This is because either they are not gape- limited (quolls) or they are partially gape-limited but so large that they are able to ingest even the biggest toad (large varanids). For these species, toad size will

175 be the most important variable determining the probability of the predator's survival after it has ingested a toad. For smaller species of gape-limited predators, however, such as most snakes and small varanid lizards, the size range of toads that can be ingested is limited by the predator's ability to swallow large prey (Arnold 1993; Shine 1991d). Thus, such predators will capture and consume only relatively small toads. For these predators, maximal toad body size within a population will be irrelevant to risk; instead, the vulnerability of such taxa will strongly depend upon the relative toxicity of the toad (i.e., its toxicity relative to body size). Thus we need to examine two traits of cane toads, variation in body size and variation in relative toxicity. Body size can be measured directly, using snout-ischium length (SIL) as a measure.

Relative toxicity is more difficult to assess but conveniently, most toxin in the skin of toads is stored in the large parotoid glands located above the shoulders

(Meyer and Linde 1971). Thus, we can use the size of the parotoids as an index of the amount of toxin carried by a toad.

176

METHODS

Collection of morphological data

I measured the 140 cane toads present in the collection of the Queensland

Museum. This specimen series represented animals collected since 1935 (the year of toad arrival). Each individual was measured for snout-ischium length

(SIL), parotoid gland length (PL) and parotoid gland width (PW). Information on collection locality and date of collection were also taken from the museum database.

Collection of climatic data

I used several climatic variables that plausibly might influence toad morphology: annual mean temperature (AMT), minimum temperature of the coldest period (MinTempCP), annual precipitation (APrecip), precipitation seasonality (PrecipSeas), moisture index seasonality (MoisIndSeas) and annual mean humidity at 1500 h (AMHumid). Each of these variables was obtained from climate layers I derived using the program ANUCLIM (Hutchinson et al.

1999) and a digital elevation model of Australia with 0.05º grid cells. Toad locality data were laid over the resultant climate grids in ARCVIEW. I extracted climatic data for each locality using the ARCVIEW extension

BIOCLIMav (Moussalli 2003).

Collection of data on time since colonisation

177

More than 2000 records of toad locality and date were available from the

Queensland Museum and from the dataset collected by Floyd et al. (1981).

These records were used to interpolate a map surface describing the arrival date of toads throughout Queensland (see Phillips and Shine submitted-a for details,

Fig. 1). Following the derivation of this surface, the Queensland Museum toad locality records were plotted and the year of toad arrival at each site was extracted. For each measured toad I subtracted the year of toad arrival (from the GIS layer) from the collection year (from the Queensland Museum database) to yield time since colonisation (TSC) – that is, the number of years a population of toads had been present in an area at the time a toad was collected.

Data analysis

My primary aim was to derive a model that would capture as much of the effect of the environment on toad morphology as possible. For this purpose, I adopted a model selection approach. By testing the relative information content of all possible models, I selected the most parsimonious model (i.e. the model that explained the most variance with the least number of factors) describing environmental variation in toad morphology.

Once I selected the best model I used it to map predicted body size and relative parotoid size within the toad’s distribution. I predicted that collection month (the month in which a toad was collected), latitude, annual mean temperature, annual precipitation, minimum temperature of the coldest period,

178 precipitation seasonality, moisture index seasonality and annual mean humidity may all influence toad morphology. Because many of the climatic variables were correlated to varying degrees, and to reduce the number of factors, I calculated the first three principal components of climatic and latitude variables. These principal components were used, along with TSC and collection month (CM) as independent variables in a multiple regression. I predicted that parotoid size and body size might vary throughout the year, hence the inclusion of CM. However if morphology does vary with CM, it is likely to be a quadratic relationship, therefore I also included a CM2 term. Two analyses were run for each species. The first used toad snout-ischium length

(SIL) as the dependent variable and the second used toad parotoid size.

Parotoid size (PS) was calculated as the first principal component of the two parotoid size variables I measured (PL and PW). The multiple regression for toad parotoid size also included toad body size (SIL) as a fixed independent variable as I was only interested in changes in relative parotoid size. I log- transformed all variables prior to the calculation of principal components.

Those variables not involved in principal components (TSC and CM) were mean-centred ( y'= y − y ) prior to analysis. Mean-centring (such that the new mean is zero) ensures that estimated coefficients are informative even in the presence of interactions; this method also reduces colinearity between variables and their interaction terms (Jaccard and Turrisi 2003).

With six independent, non-fixed variables I had 62 combinations of primary variables that could produce a model (ignoring interaction terms). To

179 make model exploration and interpretation tractable I only examined first order interactions between factors. Each of these 62 combinations was run as a full model and I deleted interaction terms if p-values indicated they were not significant (i.e., p > 0.05). For each combination of primary variables I thus derived the most parsimonious reduced model and I calculated the Akaike information criterion (AIC) value for this model. I collected the best set of models for each species and each independent variable based upon these AIC values, with models <2 units from the best model (i.e. Δi < 2) retained within the best set (Burnham and Anderson 2001). All statistical analyses were performed in JMP (v5).

Following the derivation of the best model, I reconstructed this model in

Arcview, using the component factors, to map environmental variation in toad morphology within the toad’s Queensland range. Because collection month was an important factor in both models (see results) I calculated each model for each month (1-12). I then averaged the results across the 12 months (to describe the spatial variation in morphology) and also calculated the coefficient of variation (to describe the relative amount of seasonal variation). Because the relative parotoid size model also included toad body size (SIL) as a factor I calculated all maps for the relative parotoid size of a 60mm toad.

180

Figure 1. GIS layer describing the timing (by year) of the cane toad invasion in Queensland, Australia. The extreme western edge of the distribution follows the extent of distribution records in Queensland and may not accurately reflect the actual invasion extent. Data from Floyd et al. (1981) and the specimen register of the Queensland Museum.

181

RESULTS

Calculation of multivariate model components

The first three principal components of climatic and latitude variables accounted for almost 95% of the variation in these seven factors (Table 1).

Eigenvectors indicate that PCClimLat1 is principally a latitude/temperature axis and captures most of the latitudinal variation in the environmental factors.

PCClimLat2 appears to be principally a precipitation axis, capturing the resultant variation in humidity and moisture index seasonality. PCClimLat3 is more difficult to interpret but may be capturing altitudinal variation in the environmental variables.

For toad parotoid size (PS), the first principal component of PW and PL captured more than 99% of the variation in both these variables (reflecting their strong correlation) with equal loadings on both.

Toad snout-ischium length

The best model describing spatial and temporal variation in toad body size included all six independent variables and accounted for more than 38% of the variation in toad SIL. However I was unable to exclude the possibility of a second model in which PCClimLat3 was absent (Table 2).

182

Table 1: Results of principal component analysis of climate and latitude variables for cane toad records from the Queensland Museum.

Principal components: PCClimLat1 PCClimLat2 PCClimLat3 Eigenvalue 3.96 2.40 0.27 Cumulative percent 56.57 90.88 94.68

Eigenvectors Ln(AMT) 0.463 -0.082 0.610 Ln(Aprecip) 0.130 0.612 -0.097 Ln(PrecipSeas) 0.471 -0.147 -0.393 Ln(MoisIndexSeas) 0.359 -0.400 -0.195 Ln(MinTempCP) 0.433 0.254 0.457 Ln(AMHumid) 0.088 0.610 -0.222 Ln(DecLat) -0.471 -0.017 0.410

Abbreviations are as follows: AMT, Annual mean temperature; Aprecip, Annual precipitation; PrecipSeas, Precipitation seasonality; MoisIndSeas, Moisture index seasonality; MinTempCP, Minimum temperature of the coldest period; AMHumid, Annual mean humidity at 3pm; DecLat, Decimal latitude. Eigenvector weights > 0.3 are shown in bold.

Table 2: Parameter estimates for the best models describing toad body size (SIL) and relative parotoid size as a function of climate (PCClimLat1, 2 and

FactorsSnout-ischium length Parotoid size Model 1 Model 2 Model 1 Intercept -0.183 -0.146 -0.009

SIL - - 2.529 PCClimLat1 -0.050 -0.052 0.039 PCClimLat2 0.161 0.125 0.049 PCClimLat3 -0.195 - -0.131 TSC -1.290 -1.547 -0.370 CM -0.001 0.005 0.009 CM2 0.013 0.011 -

PC1*PC2 -0.064 - -0.044 PC1*CM 0.015 0.018 - PC1*TSC - - 0.176 PC2*TSC 1.013 0.735 0.279 PC2*CM -0.031 -0.029 - PC3*CM - - 0.024

Δi 0 1.926 0 r2 0.387 0.355 0.987 3), time since colonisation (TSC) and collection month (CM).

183

Coefficients significantly different from zero are shown in bold. For brevity, PCClimLat has been further abbreviated to PC for interaction terms. Δi represents the difference in AIC value from the best model.

184

Toad parotoid size

The single best model describing variation in relative parotoid size was also complex, involving six independent variables (Table 2). All the climatic variables and both TSC and CM were present, although CM2 was absent. The single best model accounted for 98.7% of the variation in absolute parotoid size.

However, almost all of this variance (98%) was explained simply by the size variable, SIL (i.e., bigger toads have bigger parotoid glands). Thus, my model explained 0.7% of the remaining 2% of variance. Hence, factors other than SIL

(i.e. climatic and spatial factors) accounted for 35% of the residual variation in parotoid size (i.e. relative parotoid size).

Mapping spatial variation in toad size and relative parotoid size

In both cases, my models explained approximately 35% of the variation in the variable of interest (toad size or relative parotoid size). However, in both cases the model is complex, with several interaction terms making interpretation difficult. To examine the spatial pattern of body size and relative parotoid size variation, I translated my best models into GIS layers derived from the appropriate independent variables.

a) Toad body size.

Despite significant spatial variation in average toad size, seasonal variation was even greater (Fig. 2). A high coefficient of variation across most of Queensland points to strong seasonal fluctuations in average body size,

185 particularly in the north. This most likely reflects a massive, seasonal influx of young toads each year, due to seasonal recruitment. The model predicts that average body size of toads will be highest in the Wet Tropics and south-east

Queensland, with low to moderate levels of seasonal variation in those areas.

b) Relative parotoid size.

The spatial pattern in predicted relative parotoid size was complex but did not exhibit the massive seasonality apparent in the SIL model (Fig. 3). High seasonal variation was restricted to the Wet Tropics and small areas of south- east Queensland. Toads with relatively large parotoids were predicted to be present towards the north of the state with pockets of large-glanded individuals present in the Wet Tropics, mid-east and south-eastern Queensland. The model suggests that large-glanded individuals are present in some of the areas that have been occupied by toads for long periods, despite an overall trend for relative parotoid size to decrease with time since colonisation (Chapter 7 and cf.

Figs 1 and 3).

186

A)

B)

187

Figure 2. Maps of the spatial and temporal variation in toad body size (SIL), as predicted by climatic and temporal data. A) Mean SIL across months and, B) the coefficient of variation for SIL across months.

A)

B)

188

Figure 3. Maps of the spatial and temporal variation in toad relative parotoid size, as predicted by climatic and temporal data. A) Mean relative parotoid size across months and, B) the coefficient of variation for relative parotoid size across months.

189

DISCUSSION

My results show a significant effect of climate, collection month and time since colonisation on toad morphology. Because the resultant models are complex, mapping them across collection months enabled a better understanding of the results than simply examining coefficients. Additionally, mapping the models also provides a description of the spatial and temporal variation in my “key traits”. Doing so revealed that while spatial variation in toad body size (SIL) appears to be influenced by climate and time since colonisation, collection month has an overwhelming effect on predicted average

SIL. Seasonal variation in SIL is thus likely to swamp the spatial effect of climate, particularly in north and north-west Queensland. In these areas, for a predator to which toad SIL is an important factor influencing the likelihood of death by poisoning (i.e. quolls and large varanids), seasonal variation in toad size is likely to be a much more important factor than any spatial variation in

SIL.

Conversely, relative parotoid size, although influenced by collection month, appears to exhibit variation that is primarily spatial and is affected by both time since colonisation and climatic variables. For predators constrained by prey size (and thus for which relative parotoid size determines predator vulnerability, e.g. snakes and small varanids), there appears to be meaningful spatial variation in the potential impact imposed by toads. Some of this

190 variation can be explained by climatic variables and some of it can be explained by time since colonisation.

Interestingly, my models suggest that time since colonisation (TSC) has had a significant effect on toad morphology, in terms of both SIL and relative parotoid size. In areas where toads have been for a long period, the animals tend to be relatively small and to have relatively small parotoid glands. My maps of toad morphology represent an extrapolation from the original dataset, in that TSC is currently greater than 66 years in many areas although it rarely exceeded 55 years in the dataset. The extrapolation of the negative trend in relative parotoid size with TSC is reflected in predicted values of relative parotoid size that are rarely positive in my map. The strong effect of recent colonisation history on toad morphology is a powerful reminder of the importance of history in landscape-level patterns, and thus of the need to incorporate, wherever possible, recent history into landscape-level models.

Although spatiotemporal projection of my SIL model predicted high seasonal variation in toad body size, some areas are expected to maintain large average body sizes throughout the year (the Wet Tropics and south-east

Queensland). These large body size areas included parts of the state where toads were first introduced (Figs 1 and 2). Superficially, the prediction of large body sizes in some areas long after initial colonisation appears inconsistent with the overall negative effect (i.e. partial coefficient) of TSC (see also Chapter

7). Mapping also revealed that some of the areas with the largest relative parotoid size are areas where toads were first introduced (the same anomalous

191 areas as for variation in SIL, Figs 1 and 3). This is another apparent contradiction of the overall prediction of a reduction in parotoid size with increasing time since colonisation (based simply on the partial coefficient for

TSC).

These apparent anomalies can be explained by two facts. First, my model predicts shifts in the effect of TSC with precipitation and latitude (i.e. interactions). For SIL, the effect of TSC will be negative except in areas of high precipitation; for relative parotoid size both high precipitation and low latitude/high temperature will change the effect of TSC in a positive direction.

These complexities reduce and sometimes reverse overall trends of toad body size relative to time since colonisation. Second, toads were not introduced to climatically random areas of Queensland. They were first introduced into major sugar cane growing districts – i.e. areas with high precipitation and warm temperatures. Thus, the areas where toads were first introduced are the same areas where my model suggests that time since colonisation should cause little or no decrease in either body size or relative parotoid size.

My best model describing toad size accounted for 38% of the variation in toad body size. Toad age presumably accounts for a large portion of the additional variation in toad size and was not explicit in my model. For relative parotoid size, my best model accounted for 35% of the variation. Other factors such as water pH or the presence or absence of predators and competitors are known to affect amphibian morphology (Relyea 2001; van Buskirk 2002) and data on such topics were unobtainable in the current study. Further

192 investigation into the factors affecting the relative toxicity of toads would be of great interest, to further tease apart the processes generating spatial and temporal variation in morphological traits of this invading species.

Nevertheless, my models do indicate significant variation in relative parotoid size due to climate and colonisation history, and thus allow me to identify spatial variation in the likely intensity of selection that toads impose on native predators. This is especially true for gape-limited predators that are restricted to a specific size range of toads. Because of the spatial variation in relative toxicity of toads, two identical predators in different areas feeding on same-sized toads will have different chances of ingesting a fatal dose of toxin.

My model demonstrates that at least some of this difference in predator vulnerability can be related to differences in climate and time since colonisation. Further work is necessary to elucidate other environmental factors affecting toad parotoid size. An improved understanding of the factors influencing parotoid size will allow the identification of areas where impact from toads is weakest and it is in these areas that impacted native populations have the highest chance of survival and eventual adaptation.

The approach that I have adopted in this analysis is potentially applicable to many invasive species systems. Identifying traits mediating the impact on natives and then quantifying spatial variation in those traits provides a tool for predicting the level of impact and how it varies in time and space.

Both conservation biologists and managers can use such information in their attempts to mitigate impacts of invasion. Such analyses may also prove useful

193 to ecologists and evolutionary biologists utilising invasive species systems to answer theoretical questions.

ACKNOWLEDGEMENTS.

I thank Patrick Couper, Andrew Amey and Heather Janetski at the Queensland museum for access to specimens, entertaining discussion and cups of tea.

Michael Kearney provided helpful advice on analyses and along with Michael

Wall and Richard Shine, provided a constructive review of an earlier draft. I thank the Australian Research Council and Norman Wettenhall Foundation for financial support.

194

LITERATURE CITED

Agrawal, A. A., and C. N. Klein. 2000. What omnivores eat: direct effects of induced plant resistance on herbivores and indirect consequences for diet selection by omnivores. Journal of Animal Ecology 69:525-535. Alford, R. A., M. P. Cohen, M. R. Crossland, M. N. Hearnden, and L. Schwarzkopf. 1995. Population biology of Bufo marinus in northern Australia, Pages 173-181 in C. M. Finlayson, ed. Supervisong Scientist report 101: Wetland research in the wet-dry tropics of Australia. Canberra, Commonwealth of Australia. Allendorf, F. W., and L. L. Lundquist. 2003. Introduction: Population biology, evolution, and control of invasive species. Conservation Biology 17:24-30. Arnold, S. J. 1981. Behavioural variation in natural populations I: phenotypic, genetic and environmental correlations between chemoreceptive responses to prey in the garter snake, Thamnophis elegans. Evolution 35:489-509. —. 1992. Behavioral variation in natural populations - 6. Prey responses by two species of garter snakes in three regions of sympatry. Animal Behaviour 44:705-719. —. 1993. Foraging theory and prey-size-predator-size relations in snakes, Pages 87-116 in R. A. Seigel, and J. T. Collins, eds. Snakes. Ecology and behavior. New York, McGraw-Hill. Arnold, S. J., and C. R. Peterson. 1989. A test for temperature effects on the ontogeny of shape in the garter snake Thamnophis sirtalis. Physiological Zoology 62:1316-1333. Ashley, M. V., M. F. Willson, O. R. W. Pergams, D. J. O'Dowd, S. M. Gende, and J. S. Brown. 2003. Evolutionarily enlightened management. Biological Conservation 111:115-123. Ashton, K. G., and C. R. Feldman. 2003. Bergmann's rule in non-avian reptiles: turtles follow it, lizards and snakes reverse it. Evolution 57:1151-1163. Barry, F. E., P. J. Weatherhead, and D. P. Phillip. 1992. Multiple paternity in a wild population of northern water snakes, Nerodia sipedon. Behavioral Ecology and Sociobiology 30:193-199. Bernays, E., and M. Graham. 1988. On the evolution of host specificity in phytophagous arthropods. Ecology 69:886-892. Bonnet, X., D. Bradshaw, and R. Shine. 1998. Capital vs income breeding: an ectothermic perspective. Oikos 83:333-342. Bonnet, X., R. Shine, G. Naulleau, and C. Thiburce. 2001. Plastic vipers: influence of food intake on the size and shape of gaboon vipers (Bitis gabonica). Journal of Zoology, London 255:341-351. Bradshaw, W. E., and C. M. Holzapfel. 2001. Genetic shift in photoperiodic response correlated with global warming. Proceedings of the National Academy of Sciences of the United States of America 98:14509-14511. Breeden, K. 1963. Cane toad (Bufo marinus). Wildlife in Australia 1:31. Brodie, E. D., III , and E. D. Brodie, Jr. 1999a. Predator-prey arms races: asymmetrical selection on predators and prey may be reduced when prey are dangerous. BioScience 49:557-568. Brodie, E. D., III, and E. D. Brodie, Jr. 1990. Tetrodotoxin resistance in garter snakes: an evolutionary response of predators to dangerous prey. Evolution 44:651-659. —. 1999b. Costs of exploiting poisonous prey: evolutionary trade-offs in a predator- prey arms race. Evolution 53:626-631.

195

Brodie, E. D., III, and T. Garland, Jr. 1993. Quantitative genetics of snake populations, Pages 315-362 in R. A. Seigel, and J. T. Collins, eds. Snakes: Ecology and behaviour. New York, McGraw-Hill. Brodie, E. D., III, and N. H. Russell. 1999. The consistency of individual differences in behaviour: temperature effects on antipredator behaviour in garter snakes. Animal Behaviour 57:445-451. Brodie, E. D., Jr., B. J. Ridenhour, and E. D. Brodie, III. 2002. The evolutionary response of predators to dangerous prey: Hotspots and coldspots in the geographic mosaic of coevolution between garter snakes and newts. Evolution 56:2067-2082. Brown, J. S. 1999. Vigilance, patch use and habitat selection: Foraging under predation risk. Evolutionary Ecology Research 1:49-71. Burghardt, G. M., H. C. Wilcoxon, and J. A. Czaplicki. 1973. Conditioning in garter snakes: aversion to palatable prey induced by delayed illness. Animal Learning and Behaviour 1:317-320. Burnett, S. 1997. Colonizing cane toads cause population declines in native predators: reliable anecdotal information and management implications. Pacific Conservation Biology 3:65-72. Burnham, K. P., and D. R. Anderson. 2001. Kullback-Leibler information as a basis for strong inference in ecological studies. Wildlife Research 28:111-119. Calder, W. A., III. 1984, Size, function and life history. Cambridge, Harvard University Press. Catling, P. C., A. Hertog, R. J. Burt, J. C. Wombey, and R. I. Forrester. 1999. The short-term effect of cane toads (Bufo marinus) on native fauna in the Gulf Country of the Northern Territory. Wildlife Research 26:161-185. Caughley, G. 1994. Directions in conservation biology. Journal of Animal Ecology 63:215-244. Chapin, F. S., E. S. Zavaleta, V. T. Eviner, R. L. Naylor, P. M. Vitousek, H. L. Reynolds, D. U. Hooper et al. 2000. Consequences of changing biodiversity. Nature 405:234-242. Chen, K. K., and A. Kovarikova. 1967. Pharmacology and toxicology of toad venom. Journal of Pharmaceutical Sciences 56:1535-1541. Christians, J. K. 2002. Avian egg size: variation within species and inflexibility within individuals. Biological Reviews 77:1-26. Cogger, H., E. Cameron, R. Sadlier, and P. Eggler. 1993, The action plan for australian reptiles. Canberra, Australian Nature Conservation Agency. Cogger, H., and H. Heatwole. 1981. The Australian reptiles: Origins, biogeography, distribution patterns and island evolution, Pages 1333-1373 in A. Keast, ed. Ecological biogeography of Australia. The Hague, Junk. Cogger, H. G. 2000, Reptiles and amphibians of Australia. Melbourne, Reed Books. Covacevich, J., and M. Archer. 1975. The distribution of the cane toad, Bufo marinus, in Australia and its effects on indigenous vertebrates. Memoirs of the Queensland Museum 17:305-310. Crossland, M. R. 1998. Ontogenetic variation in toxicity of tadpoles of the introduced toad Bufo marinus to native Australian aquatic invertebrate predators. Herpetologica 54:364-369. —. 2000. Direct and indirect effects of the introduced toad Bufo marinus (Anura: Bufonidae) on populations of native anuran larvae in Australia. Ecography 23:283-290.

196

—. 2001. Ability of predatory native Australian fishes to learn to avoid toxic larvae of the introduced toad Bufo marinus. Journal of Fish Biology 59:319-329. Crossland, M. R., and R. A. Alford. 1998. Evaluation of the toxicity of eggs, hatchlings and tadpoles of the introduced toad Bufo marinus (Anura, Bufonidae) to native Australian aquatic predators. Australian Journal of Ecology 23:129-137. Crossland, M. R., and C. Azevedo-Ramos. 1999. Effects of Bufo (Anura : Bufonidae) toxins on tadpoles from native and exotic Bufo habitats. Herpetologica 55:192- 199. Cundall, D., and H. W. Greene. 2000. Feeding in snakes, Pages 293-333 in K. Schwenk, ed. Feeding. Form, function and evolution in tetrapod vertebrates. San Diego, Academic Press. D'Antonio, C. M., and S. Kark. 2002. Impacts and extent of biotic invasions in terrestrial ecosystems. Trends in Ecology & Evolution 17:202-204. Daly, J. W., C. W. Myers, and N. Whittaker. 1987. Further classification of skin alkaloids from neotropical poison frogs (Dendrobatidae) with a general survey of toxic/noxious substances in the amphibia. Toxicon 25:1023-1095. Daly, J. W., and B. Witkop. 1971. Chemistry and pharmacology of frog venoms, Pages 497-519 in W. Bucherl, and E. E. Buckley, eds. Venomous animals and their venoms. New York, Academic Press. Demott, W. R., and F. Moxter. 1991. Foraging on cyanobacteria by copepods - responses to chemical defenses and resource abundance. Ecology 72:1820-1834. Diamond, J. M. 1989. Overview of recent extinctions, Pages 37-41 in D. Western, and M. C. Pearl, eds. Conservation for the 21st century. Oxford, Oxford University Press. Dicke, M. 2000. Chemical ecology of host-plant selection by herbivorous arthropods: a multitrophic perspective. Biochemical Systematics and Ecology 28:601-617. Downes, S. 2001. Trading heat and food for safety: Costs of predator avoidance in a lizard. Ecology 82:2870-2881. Duellman, W. E., and L. Trueb. 1994, Biology of amphibians. Baltimore MD, John Hopkins University Press. Easteal, S. 1981. The history of introductions of Bufo marinus (Amphibia: Anura); a natural experiment in evolution. Biological Journal of the Linnean Society 16:93-113. Easteal, S., E. K. Van Beurden, R. B. Floyd, and M. D. Sabath. 1985. Continuing geographical spread of Bufo marinus in Australia: a range expansion between 1974 and 1980. Journal of Herpetology 19:185-188. Ehrlich, P. R. 1995. The scale of human enterprise and biodiversity loss, Pages 214-226 in J. H. Lawton, and R. M. May, eds. Extinction rates. Oxford, Oxford University Press. Elton, C. S. 1958, The ecology of invasions by animals and plants. London, Methuen. Erspamer, V., G. F. Erspamer, G. Mazzanti, and R. Endean. 1984. Active peptides in the skins of one hundred amphibian species from Australia and Papua New Guinea. Comparative Biochemistry & Physiology - C: Comparative Pharmacology & Toxicology 77:99-108. Erspamer, V., M. Roseghini, R. Endean, and A. Anastasi. 1966. Biogenic amines and active polypeptides in the skin of Australian amphibians. Nature 212:204. Estoup, A., I. J. Wilson, C. Sullivan, J. M. Cornuet, and C. Moritz. 2001. Inferring population history from microsatellite and enzyme data in serially introduced cane toads, Bufo marinus. Genetics 159:1671-1687.

197

Ewald, P. W. 1994, Evolution of infectious disease. New York, Oxford University Press. Falconer, D. S., and T. F. C. Mackay. 1996, Introduction to quantitative genetics. Burnt Mill, Harlow, England, Longman. Fearn, S. 2003. Pseudechis porphyriacus (Red-bellied black snake). Diet. Herpetological Review 34:253-254. Flannery, T. F. 1994, The future eaters: An ecological history of the Australasian lands and people. Chatswood, NSW, Reed. Flier, J., M. W. Edwards, J. W. Daley, and C. W. Myers. 1980. Widespread occurrence in frogs and toads of skin compounds interacting with the ouabain site of Na+, K+-ATPase. Science 208:503-505. Floyd, R. B., W. C. Boughton, S. Easteal, M. D. Sabath, and E. K. van Beurden. 1981, The distribution records of the marine toad (Bufo marinus). 1. Australia.: AES Working Paper 3/81. Brisbane, Griffith University. Forsman, A. 1996. An experimental test for food effects on head size allometry in juvenile snakes. Evolution 50:2536-2542. Freeland, W. J. 1986. Populations of cane toad Bufo marinus in relation to time since colonization. Australian Wildlife Research 13:321-330. —. 1987. Cane toads and the balance of nature. Wildlife Australia:12-15. Freeland, W. J., and S. H. Kerin. 1990. Within habitat relationships between invading Bufo marinus and Australian species of frog during the tropical dry season. Australian Wildlife Research 15:293-305. Fritts, T. H., and G. H. Rodda. 1998. The role of introduced species in the degredation of island ecosystems: a case history of Guam. Annual Review of Ecology and Systematics 29:113-140. Gardner, M. J., N. Hall, E. Fung, O. White, M. Berriman, R. W. Hyman, J. M. Carlton et al. 2002. Genome sequence of the human malaria parasite Plasmodium falciparum. Nature 419:498-511. Gilchrist, G. W., R. B. Huey, J. Balanya, M. Pascual, and L. Serra. 2004. A time series of evolution in action: A latitudinal cline in wing size in South American Drosophila subobscura. Evolution 58:768-780. Gomulkiewicz, R., J. N. Thompson, R. D. Holt, S. L. Nuismer, and M. E. Hochberg. 2000. Hot spots, cold spots and the geographic mosaic theory of coevolution. The American Naturalist 156:157-174. Goodacre, W. A. 1947. The giant toad (Bufo marinus); an enemy of bees. Agricultural Gazette of New South Wales 58:374-375. Gould, S. J. 1966. Allometry and size in ontogeny and phylogeny. Biological Reviews of the Cambridge Philosophical Society 41:587-640. —. 1991, Bully for Brontosaurus. New York, Norton. Grant, B. S. 1999. Fine tuning the peppered moth paradigm. Evolution 53:980-984. Grant, P. R., and B. R. Grant. 2002. Unpredictable evolution in a 30-year study of Darwin's finches. Science 296:707-711. Greene, H. W. 1997, Snakes. The evolution of mystery in nature. Berkeley, University of California Press. Greer, A. E. 1997, The biology and evolution of Australian snakes. Chipping Norton, Surrey Beatty and Sons. Grosholz, E. D., and G. M. Ruiz. 2003. Biological invasions drive size increases in marine and estuarine invertebrates. Ecology Letters 6:700-705.

198

Guglielmo, C. G., W. H. Karasov, and W. J. Jakubas. 1996. Nutritional costs of a plant secondary metabolite explain selective foraging by ruffed grouse. Ecology 77:1103-1115. Hendry, A. P., and M. T. Kinnison. 1999. The pace of modern life: Measuring rates of contemporary microevolution. Evolution 53:1637-1653. Hewitt, G. C. 1956. The giant American toad. Walkabout 22:45. Higgie, M., S. Chenoweth, and M. W. Blows. 2000. Natural selection and the reinforcement of mate recognition. Science 290:519-521. Hoffmann, A. A., R. J. Hallas, J. A. Dean, and M. Schiffer. 2003. Low potential for climatic stress adaptation in a rainforest Drosophila species. Science 301:100- 102. Houston, A. I., J. M. McNamara, and J. M. C. Hutchinson. 1993. General results concerning the trade-off between gaining energy and avoiding predation. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 341:375-397. Huey, R. B., G. W. Gilchrist, M. L. Carlson, D. Berrigan, and L. Serra. 2000. Rapid evolution of a geographic cline in size in an introduced fly. Science 287:308- 309. Hutchinson, M. F., D. J. Houlder, H. A. Nix, and J. P. McMahon. 1999, ANUCLIM user guide, version 5.1. Canberra, Centre for Resource and Environmental Studies, Australian National University. Ingram, G. J., and J. Covacevich. 1990. Tropidonophis mairii vs Bufo marinus. Memoirs of the Queensland Museum 29:396. IUCN. 2001. 100 of the world's worst invasive species: a selection from the global invasive species database. http://www.issg.org/booklet.pdf, IUCN. Jaccard, J., and R. Turrisi. 2003, Interaction effects in multiple regression: Sage University Paper Series on Quantitative Applications in the Social Sciences, 07- 72. Thousand Oaks, CA, Sage Publications. Kettlewell, B. 1973, The evolution of melanism. Oxford, Clarendon Press. Kiesecker, J. M., and A. M. Blaustein. 1997. Population differences in responses of red- legged frogs (Rana aurora) to introduced bullfrogs. Ecology 78:1752-1760. Kinghorn, J. R. 1938. The giant toad Bufo marinus in Australia. The Australian Museum Magazine 6:404-411. Kolbe, J. J., R. E. Glor, L. R. Schettino, A. C. Lara, A. Larson, and J. B. Losos. 2004. Genetic variation increases during biological invasion by a Cuban lizard. Nature 431:177-181. Lawler, K. L., and J. M. Hero. 1997. Palatability of Bufo marinus tadpoles to a predatory fish decreases with development. Wildlife Research 24:327-334. Lee, C. E. 2002. Evolutionary genetics of invasive species. Trends in Ecology & Evolution 17:386-391. Lever, C. 2001, The cane toad. The history and ecology of a successful colonist. Yorkshire, Westbury Academic and Scientific Publishing. Losos, J. B., K. I. Warheit, and T. W. Schoener. 1997. Adaptive differentiation following experimental island colonization in Anolis lizards. Nature 387:70-73. Low, T. 1999, Feral future. Ringwood, Victoria, Viking. Lutz, B. 1971. Venomous toads and frogs, Pages 423-473 in W. Bucherl, and E. E. Buckley, eds. Venomous animals and their venoms. New York, Academic Press. Lynch, M., and B. Walsh. 1998, Genetics and Analysis of Quantitative Traits. Sunderland, MA, Sinauer Associates.

199

Mack, R. N., D. Simberloff, W. M. Lonsdale, H. Evans, M. Clout, and F. Bazzaz. 2000. Biotic invasions: Causes, epidemiology, global consequences and control. Issues in Ecology 5:1-20. Macnair, M. 1987. Heavy metal tolerance in plants: a model evolutionary system. Trends in Ecology & Evolution 2:354-359. Madsen, T., and R. Shine. 1994. Toxicity of a tropical Australian frog, Litoria dahlii, to sympatric snakes. Wildlife Research 21:21-25. Malnate, E. V., and G. Underwood. 1988. Australasian snakes of the genus Tropidonophis. Proceedings of the Academy of Natural Sciences Philadelphia 140:59-201. Manly, B. F. J. 1991, Randomization and monte carlo methods in biology. Melbourne: Australia, Chapman and Hall. May, R. M., J. H. Lawton, and N. M. Stork. 1995. Assessing extinction rates, Pages 1- 24 in J. H. Lawton, and R. M. May, eds. Extinction rates. Oxford, Oxford University Press. McCracken, G. F., G. M. Burghardt, and S. E. Houts. 1999. Microsatellite markers and multiple paternity in the garter snake Thamnophis sirtalis. Molecular Ecology 8:1475-1479. McDowell, S. B. 1972. The species of Stegonotus (serpentes, colubridae) in Papua New Guinea. Zoologische Mededelingen 47:6-26. Merila, J., B. C. Sheldon, and L. E. B. Kruuk. 2001. Explaining stasis: microevolutionary studies in natural populations. Genetica 112:199-222. Meyer, K., and H. Linde. 1971. Collection of toad venoms and chemistry of toad venom steroids, Pages 521-556 in W. Bucherl, and E. E. Buckley, eds. Venomous animals and their venoms. New York, Academic Press. Miller, D. E., and H. R. Mushinsky. 1990. Foraging ecology and prey size in the mangrove water snake, Nerodia fasciata compressicauda. Copeia 1990:1099- 1106. Morrison, R., and M. Morrison. 1991, The voyage of the great southern ark. The 4 billion year journey of the Australian continent. Sydney, Ure Smith Press. Motychak, J. E., E. D. Brodie, Jr., and E. D. Brodie, III. 1999. Evolutionary response of predators to dangerous prey: preadaptation and the evolution of tetrodotoxin resistance in garter snakes. Evolution 53:1528-1535. Moussalli, A. 2003, BIOCLIMav v1.0, user guide. Brisbane, Cooperative Research Centre for Tropical Rainforest Ecology and Management. Mousseau, T. A. 1997. Ectotherms follow the converse to Bergmann's rule. Evolution 51:630-632. Mungomery, R. W. 1935. The giant american toad (Bufo marinus). Cane Grower's Quarterly Bulletin 3:21-27. Novacek, M. J., and E. E. Cleland. 2001. The current biodiversity extinction event: Scenarios for mitigation and recovery. Proceedings of the National Academy of Sciences of the United States of America 98:5466-5470. Oakwood, M. 2003. The effect of cane toads on a marsupial carnivore, the northern quoll, Dasyurus hallucatus, Parks Australia North. Ogutu-Ohwayo, R. 1999. Nile perch in Lake Victoria: the balance between benefits and negative impacts of aliens, Pages 47-64 in O. T. Sandlund, P. J. Schei, and A. Viken, eds. Invasive species and biodiversity management. Boston, Kluwer Academic.

200

Olsen, E. M., M. Heino, G. R. Lilly, M. J. Morgan, J. Brattey, B. Ernande, and U. Dieckmann. 2004. Maturation trends indicative of rapid evolution preceded the collapse of northern cod. Nature 428:932-935. Palumbi, S. R. 2002, The evolution explosion: How humans cause rapid evolutionary change. New York, Norton. Parker, I. M., J. Rodriguez, and M. E. Loik. 2003. An evolutionary approach to understanding the biology of invasions: Local adaptation and general-purpose genotypes in the weed Verbascum thapsus. Conservation Biology 17:59-72. Perry, G., and E. R. Pianka. 1997. Animal foraging: past, present and future. Trends in Ecology & Evolution 12:360-364. Phillips, B. L., G. P. Brown, and R. Shine. 2003. Assessing the potential impact of cane toads on Australian snakes. Conservation Biology 17:1738-1747. Phillips, B. L., and M. Fitzgerald. 2004. Encounters between eastern brown snakes (Pseudonaja textilis) and cane toads (Bufo marinus) in northern New South Wales. Herpetofauna 34:23-25. Phillips, B. L., and R. Shine. submitted-a. Adapting to an invasive species: toxic cane toads induce morphological change in Australian snakes. —. submitted-b. Allometry and selection in a novel predator/prey system: Australian snakes and the invading cane toad. Pockley, D. 1965. The free and the caged. Blackwoods Magazine 298:439-446. Pope, C. H. 1935, The Reptiles of China: Natural History of Central Asia, v. 10. New York, American Museum of Natural History. Pough, F. H. 1980. The advantages of ectothermy for tetropods. The American Naturalist 115:92-112. Queraz-Regal, A., and R. B. King. 1998. Evidence for phenotypic plasticity in snake body size and relative head dimensions in response to amount and size of prey. Copeia 1998:423-429. Raup, D. M. 1993, Extinction: bad genes or bad luck? Oxford, Oxford University Press. Rausher, M. D. 2001. Co-evolution and plant resistance to natural enemies. Nature 411:857-864. Rayward, A. 1974. Giant toads - a threat to Australian wildlife. Wildlife 17:506-507. Relyea, R. A. 2001. The lasting effects of adaptive plasticity: predator-induced tadpoles become long-legged frogs. Ecology 82:1947-1955. Reznick, D., F. Shaw, F. Rodd, and R. Shaw. 1996. Evaluation of the rate of evolution in natural populations of guppies (Poecilia reticulata). Science 275:1934-1937. Roff, D. A. 1997, Evolutionary quantitative genetics. New York, Chapman and Hall. Sabath, M. D., W. C. Boughton, and S. Easteal. 1981. Expansion of the range of the introduced toad Bufo marinus in Australia 1935-1974. Copeia 1981:676-680. Savidge, J. A. 1988. Food habits of Boiga irregularis, an introduced predator on Guam. Journal of Herpetology 22:275-282. Schmidt-Nielsen, K. 1984, Scaling: why is animal size so important? Cambridge, Cambridge University Press. Schoener, T. W. 1971. Theory of feeding strategies. Annual Review of Ecology & Systematics 2:369-404. Shaw, R. G. 1987. Maximum-likelihood approaches applied to quantitative genetics of natural populations. Evolution 41:812-826. Shine, R. 1978. Growth rates and sexual maturation in six species of Australian Elapid snakes. Herpetologica 34:73-79. —. 1989. Constraints allometry and adaptation: food habits and reproductive biology of Australian brown snakes Pseudonaja Elapidae. Herpetologica 45:195-207.

201

—. 1991a, Australian snakes: a natural history. Sydney, Reed Books. —. 1991b. Intersexual dietary divergence and the evolution of sexual dimorphism in snakes. The American Naturalist 138:103-122. —. 1991c. Strangers in a strange land: ecology of the Australian colubrid snakes. Copeia 1991:120-131. —. 1991d. Why do larger snakes eat larger prey items? Functional Ecology 5:493-502. —. 1994. Sexual size dimorphism in snakes revisited. Copeia 1994:326-346. Simberloff, D., T. Dayan, C. Jones, and G. Ogura. 2000. Character displacement and release in the small Indian mongoose, Herpestes javanicus. Ecology 81:2086- 2099. Simons, A. M. 2003. Invasive aliens and sampling bias. Ecology Letters 6:278-280. Sinervo, B., and P. Doughty. 1996. Interactive effects of offspring size and timing of reproduction on offspring reproduction: Experimental, maternal, and quantitative genetic aspects. Evolution 50:1314-1327. Smith, J. G., and B. L. Phillips. submitted. Assessing the potential impact of cane toads on Australia's reptiles. Sokal, R. R., and F. J. Rohlf. 1995, Biometry: The principles and practice of statistics in biological research. New York, W.H. Freeman and Company. Stearns, S. C. 1992, The evolution of life histories. Oxford, Oxford University Press. Stockwell, C. A., A. P. Hendry, and M. T. Kinnison. 2003. Contemporary evolution meets conservation biology. Trends in Ecology & Evolution 18:94-101. Sutherst, R. W., R. B. Floyd, and G. F. Maywald. 1995. The potential geographical distribution of the cane toad, Bufo marinus L. in Australia. Conservation Biology 10:294-299. Terrick, T. D., R. L. Mumme, and G. M. Burghardt. 1995. Aposematic coloration enhances chemosensory recognition of noxious prey in the garter snake Thamnophis radix. Animal Behaviour 49:857-866. Thomas, C. D., A. Cameron, R. E. Green, M. Bakkenes, L. J. Beaumont, Y. C. Collingham, B. F. N. Erasmus et al. 2004. Extinction risk from climate change. Nature 427:145-148. Thompson, J. N. 1998. Rapid evolution as an ecological process. Trends in Ecology & Evolution 13:329-332. van Buskirk, J. 2002. Phenotypic lability and the evolution of predator-induced plasticity in tadpoles. Evolution 56:361–370. van Dam, R. A., D. J. Walden, and G. W. Begg. 2002. A preliminary risk assessment of cane toads in Kakadu National Park. Scientist Report 164. Darwin, NT, Supervising Scientist. Webb, J. K., R. Shine, and K. A. Christian. in press. Does intraspecific niche partitioning in a native predator influence its response to an invasion by a toxic prey species? Austral Ecology. Williams, C. R., E. D. Brodie, Jr., M. J. Tyler, and S. J. Walker. 2000. Antipredator mechanisms of Australian frogs. Journal of Herpetology 34:431-443. Williamson, I. 1999a. Competition between the larvae of the introduced cane toad Bufo marinus (Anura: Bufonidae) and native anurans from the Darling Downs area of southern Queensland. Australian Journal of Ecology 24:636-643. Williamson, M. 1996, Biological invasions: Population and Community Biology Series, v. 15. London, Chapman and Hall. —. 1999b. Invasions. Ecography 22:5-12. Zug, G. R., and P. B. Zug. 1979. The marine toad, Bufo marinus: a natural history resume of native populations. Smithsonian Contributions to Zoology 284.

202