Biol Invasions (2018) 20:1833–1847 https://doi.org/10.1007/s10530-018-1665-8 OR IGINAL PAPER Invasive cane toads might initiate cascades of direct and indirect effects in a terrestrial ecosystem . . . . Benjamin Feit Christopher E. Gordon Jonathan K. Webb Tim S. Jessop . . Shawn W. Laffan Tim Dempster Mike Letnic Received: 9 April 2016 / Accepted: 12 January 2018 / Published online: 19 January 2018 The Author(s) 2018. This article is an open access publication Abstract Understanding the impacts that invasive remains scarce. Here, we ask whether the invasion of vertebrates have on terrestrial ecosystems extends the cane toad, a vertebrate invader that is toxic to many primarily to invaders’ impacts on species with which of Australia’s vertebrate predators, has induced eco- they interact directly through mechanisms such as logical cascades in a semi-arid rangeland. We com- predation, competition and habitat modiﬁcation. In pared activity of a large predatory lizard, the sand- addition to direct effects, invaders can also initiate goanna, and abundances of smaller lizards preyed ecological cascades via indirect population level upon by goannas in areas of high toad activity near effects on species with which they do not directly toads’ dry season refuges and areas of low toad interact. However, evidence that invasive vertebrates activity distant from toads’ dry season refuges. initiate ecological cascades in terrestrial ecosystems Consistent with the hypothesis that toad invasion has led to declines of native predators susceptible to poisoning, goanna activity was lower in areas of high toad activity. Consistent with the hypothesis that toad- induced goanna decline lead to increases in abundance Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10530-018-1665-8) con- tains supplementary material, which is available to authorized users. B. Feit (&) S. W. Laffan M. Letnic J. K. Webb Centre for Ecosystem Science, University of New South School of the Environment, University of Technology Wales, Sydney, NSW 2052, Australia Sydney, Sydney, NSW 2007, Australia e-mail: firstname.lastname@example.org T. S. Jessop B. Feit School of Life and Environmental Sciences, Deakin Hawkesbury Institute for the Environment, Western University, Burwood, VIC 3125, Australia Sydney University, Sydney, NSW 2751, Australia T. Dempster B. Feit Department of Zoology, University of Melbourne, Department of Ecology, Swedish University of Melbourne, VIC 3010, Australia Agricultural Sciences, 75007 Uppsala, Sweden C. E. Gordon Centre for Environmental Risk Management of Bushﬁres, University of Wollongong, Wollongong, NSW 2522, Australia 123 1834 B. Feit et al. the prey of goannas, smaller lizards were more (Estes et al. 2011; Letnic et al. 2009). Ecological abundant in areas of high toad activity. Structural cascades initiated by invasive species is a relatively equation modelling showed a positive correlation well documented phenomenon in aquatic systems (e.g. between goanna activity and distance from dry season Simon and Townsend 2003; Baxter et al. 2004; Strayer refuge habitats used by toads. The abundances of small 2010) with increasing evidence suggesting similar lizards was correlated negatively with goanna activity effects of invasive vertebrates on the terrestrial and distance from dry season refuges of toads. Our ecosystems of offshore islands (Roemer et al. 2002; ﬁndings provide support for the notion that invasions Croll et al. 2005; Thoresen et al. 2017), but few studies by terrestrial vertebrates can trigger ecological have reported invasive species driving cascades in cascades. terrestrial ecosystems of mainland continents. One reason for the scarcity of studies on the indirect Keywords Cane toad Invasive species Rhinella impacts of vertebrate invaders in terrestrial ecosys- marina Semi-arid Trophic cascade Varanus tems is that studies investigating biological invasions gouldii are difﬁcult to plan. Hence, most studies reporting impacts of biological invasions are conducted post- invasion and typically evaluate the impacts of invasive vertebrates by manipulating their abundance or access Introduction to the species or ecosystem of interest. However, demonstrating that vertebrate invaders can have The invasion of ecosystems by non-indigenous plant cascading effects in terrestrial ecosystems often and animal species is a major driver of global requires conducting manipulative experiments at large environmental change and recognized as a serious spatial scales, which are logistically difﬁcult (Parker threat to biodiversity (Vitousek and D’Antonio 1997; et al. 1999). One way to advance knowledge of Mack et al. 2000). Research on the ecological invasive vertebrates’ indirect impacts on ecosystems consequences of biological invasions has focused is to utilize ‘‘natural experiments’’ whereby the largely on direct population-level impacts on single or abundance of invaders varies in time or space in speciﬁc guilds of native species through mechanisms otherwise similar landscapes. Such studies can such as predation (Risbey et al. 2000; Roy et al. 2012), provide valuable insights into ecological processes at competition (Corbin and D’Antonio 2004; Miller and spatial and temporal scales that cannot be achieved Gorchov 2004) and habitat modiﬁcation (Rodriguez through experimentation (Sax et al. 2005). In the semi- 2006). However, population-level effects that invaders arid rangelands of northern Australia, spatial structur- have on species they do not directly interact with (i.e., ing of cane toad populations, whereby their activity is indirect effects) are often overlooked, even though concentrated around dry season refuge habitats (Let- there is evidence that such effects can manifest in the nic et al. 2014, 2015), provides the opportunity to reorganization of recipient ecosystems (White et al. conduct a ‘‘large-scale’’ natural experiment to exam- 2006). ine the role that introduced species have in structuring Invaders are likely to induce cascades of indirect ecosystems. effects if they inﬂuence the abundance of strongly The cane toad (Rhinella marina) is an anuran native interactive species such as predators, pollinators and to South America, that is currently invading northern ecosystem engineers or substantially alter primary and semi-arid regions of Australia (Florance et al. productivity or vegetation structure (Anderson and 2011). Cane toads contain toxins that are absent from Rosemond 2007; Gooden et al. 2009; Letnic et al. Australian anurans. Consequently, most native Aus- 2009). In such circumstances, invaders can propagate tralian vertebrate predator species lack an evolution- ecological cascades whereby shifts in the abundance ary history of exposure to these toxins and many die of one species indirectly affect the abundance and after attacking or consuming toads (Shine 2010). Due biomass of others (Terborgh and Estes 2010). Such to poisoning of individuals, populations of marsupial ecological cascades may become evident temporally quolls, monitor lizards (i.e., goannas), freshwater or spatially as alternating patterns in the abundances of crocodiles and some snake species have undergone species are affected directly and indirectly by invaders marked declines following the arrival of toads (Shine 123 Invasive cane toads might initiate cascades 1835 2010; Feit and Letnic 2015). Because cane toads have encounter rates between toads and varanid lizards driven declines in predator populations, it follows that should decrease with distance from AWP, varanid diminished levels of predation could lead to increases lizard foraging activity should increase with distance in abundance or survivorship of prey species in areas from AWP and should be greater in the vicinity of where cane toads have suppressed the abundances of tanks where toads are rare than dams where toads are predators (Doody et al. 2013, 2015). abundant when distance from water is set to the mean Cane toads are now spreading through Australia’s distance; and (3) because of reduced predation by vast semi-arid rangelands (Tingley et al. 2014). In varanid lizards, the abundance of small lizards (Scin- semi-arid regions, cane toads require regular access to cidae and Agamidae) should be greater in the vicinity water to survive sustained periods of hot, dry weather of dams where toads are abundant than tanks where without accessing water (Florance et al. 2011; Jessop toads are rare. et al. 2013b). As a result, during prolonged periods of We tested our predictions by comparing the abun- dry conditions the distribution of cane toads in semi- dance of cane toads, the foraging activity of the sand arid landscapes is restricted to isolated populations at goanna (Varanus gouldii) (Fig. 1e, f) and the abun- places where permanent water is available (Letnic dance of skinks and dragons along 12 km road et al. 2014). Because natural sources of water are transects in the vicinity of dams and tanks, respec- normally scarce in Australia’s semi-arid regions, their tively. We used structural equation modelling (SEM) to investigate the hypothesized direct and indirect invasion has been facilitated by the presence of earthen dams at artiﬁcial water points (AWP) (Flo- relationships among the response variables. We also rance et al. 2011) (Fig. 1a). These dams function as tested alternative hypotheses in our SEM based on reservoirs for water pumped from bores and normally prior knowledge that the abundances of goannas and provide drinking water to livestock via a gravity fed smaller lizards are inﬂuenced by predation from trough (Fig. 1d). Although toads cannot normally mammalian predators (Olsson et al. 2005) and distur- access the water held in livestock troughs, water stored bances to vegetation by livestock grazing (James et al. in dams is readily accessible to them. Consequently, 1999) and ﬁre (Letnic et al. 2004). during dry periods, earthen dams function as refuges that support dense cane toad populations (Florance et al. 2011; Letnic et al. 2014) (Fig. 1b). Previous Materials and methods studies have demonstrated that cane toad population can be suppressed by restricting their access to water at Study area and time AWP either by installing toad-proof fences at dams or by using tanks made of plastic or steel as reservoirs We conducted our surveys on two neighboring cattle 0 0 instead of earthen dams (Letnic et al. 2014; Feit et al. stations, Dungowan (1642 S, 13216 E) and Camﬁeld 0 0 2015) (Fig. 1c). In rangeland areas of the Tanami (172 S, 13117 E), located in the northern margin of Desert in Australia’s Northern Territory, the existence the Tanami Desert in the Northern Territory, Aus- of AWP ﬁtted with reservoirs that support high density tralia. Bore-fed reservoirs at AWP on both stations (dams) and low density (tanks) cane toad populations consist of a mix of earthen dams and tanks made of provided us with the opportunity to conduct a large- plastic or steel (Fig. 2). At both reservoir types, scale natural experiment to examine the effects that livestock are supplied with water through troughs toads have had on lizard assemblages. located within 50 m of the reservoir (Fig. 1d). The Based on prior knowledge of cane toad biology, troughs are fed by gravity and ﬁtted with a ﬂoat-valve cane toads’ suppressive effects on varanid lizard to prevent them from over-ﬂowing. Permanent fences populations (Doody et al. 2009, 2014, 2015) and prevent livestock from accessing the water stored in varanid lizards’ predatory effects on populations of dams. smaller lizard species (Olsson et al. 2005; Doody et al. The study area has a mean annual rainfall of 2012, 2015), we tested the following predictions: (1) 580 mm, of which 96% falls in the wet season toad activity should decrease with distance from AWP (November to April) and 4% in the dry season and be greater in the vicinity of dams where toads are (May–October; Australian Bureau of Meteorology). abundant than tanks where toads are rare; (2) because The vegetation of the study area consists of open semi- 123 1836 B. Feit et al. 123 Invasive cane toads might initiate cascades 1837 bFig. 1 a Earthen dam (arrowed) used as reservoir at an AWP. Cane toad abundance at AWP and along road b Dams support large numbers of cane toads (Rhinella marina) transects because they allow ready access to water for rehydration and reproduction. c Tank used as reservoir at an AWP. In We estimated the abundance of toads in the direct comparison to dams, tanks support few toads because they allow only little access to water. d A trough from which vicinity of AWP by conducting nocturnal livestock drink in our study area. Gravity supplies the trough 4m 9 150 m strip transects radiating away from the with water from a dam or tank. Toads cannot normally access AWP (n = 4 per AWP) using handheld 12 V spot- water in troughs. e Sand goanna (Varanus gouldii). f Ellipsoid lights with 25 W halogen bulbs. Cane toad abundance foraging pit created by a sand goanna when digging for fossorial prey was calculated as the sum of individuals encountered along the four transects. We conducted a total of 42 Fig. 2 Map of the study area showing the location of transects (solid lines) in the vicinity of AWP ﬁtted with dams (black) and tanks (dark grey). The inset shows the location of study area (shaded) within Australia arid savannah woodland with the dominant woody cane toad counts at 31 AWP (ten dams and 21 tanks) species of lancewood (Acacia shirleyi) and eucalypts (Online Resource 1); four dams and seven tanks were (Eucalyptus leucophloia) and an understory domi- sampled twice during the study period with a nated by grasses (Eriachne spp. and Sorghum spp.). minimum of 12 months between surveys. We surveyed the foraging activity of sand goannas and Cane toads frequently travel along roads during the abundance of cane toads, skinks and dragons in dispersal periods (Brown et al. 2006). To document April and November 2012, April and November 2013 the distribution of toads with respect to distance from and September 2014. AWP, we conducted nocturnal surveys along low-use single lane dirt roads in a 4WD vehicle during a period 123 1838 B. Feit et al. when many toads had dispersed away from their dry survey. Track plots were spaced 500 m apart and season refuges at the end of the wet season in April located between 0 and 12 km from the nearest AWP. 2012. Because of logistical constraints during ﬁeld We walked along each track plot and recorded the work, it was not possible to conduct road transects presence or absence of fresh goanna tracks (i.e., from all AWP. Hence, distance mediated effects were distinctive tail drags and claw imprints). As daily evaluated at a subset of dams (n = 4) and tanks activity areas of sand goannas are unlikely to exceed (n = 5). We surveyed toad abundance over a total of an area of 200 m by 200 m (Green and King 1978), we 110 km at distances of up to 12 km from both are conﬁdent that each recorded track originated from reservoir types. The surveys were undertaken at a a different individual. speed of 20 km/h and an observer noted with a GPS The foraging-pit based index of goanna activity was the location of all toads sighted. Toad activity was derived by scoring the presence or absence of recent documented as number of toads per 500 m transect goanna foraging signs during 2 min active searches in section. the vicinity of each track plot (Jessop et al. 2013a). Whilst digging for fossorial prey, sand goannas leave Goanna activity indices characteristic ellipsoid foraging pits that often show deep scratch marks left by their strong forelimbs Sand goannas are difﬁcult to survey using mark- during excavation of the soil (Read and Scoleri 2014). We estimated the approximate age of foraging pits recapture methods because they rarely enter traps (Letnic et al. 2004) and, in habitats with dense based on two criteria: the amount of leaf litter and understory vegetation such as our study area, are other debris in the excavation and the coloration and difﬁcult to sight and approach for the purposes of texture of the excavated soil (initially darker and softer noosing, hand capture or visual surveys. Previous than the topsoil, gradually fading and hardening over studies have used the occurrence of fresh goanna the course of several weeks). To provide an indication tracks and pits that goannas create whilst foraging to of the age of foraging pits that we encountered, we index goanna abundance (Paltridge 2002; Bird et al. excavated pits similar to goanna foraging pits and 2014; Read and Scoleri 2014). Both indices have been monitored them over a 2 month period. This allowed validated against known abundances in other varanid us to classify foraging pits into the two age classes of species (Anson et al. 2014). Following these previous recent (i.e. younger than approximately 1 month) and studies, we used two methods to index goanna activity, old (i.e. older than 1 month). Only the presence of the occurrence of tracks (i.e. footprints and tail drag foraging pits younger than approximately 1 month marks) crossing single-lane dirt roads and the occur- was used for further analyses. rence of recent goanna foraging pits. Our track count index provided a measurement of goanna activity over Small lizard abundance a 24 h period, while the foraging pit index provided a cumulative measure of goanna activity for a period of To monitor the abundance of small lizards, we approximately 1 month prior to our surveys. We conducted 198 active diurnal searches (total search conducted all monitoring under environmental condi- duration of 1980 min) following the methods of tions that favored lizard activity and ensured equal and Lunney and Barker (1986) (Online Resource 3). high detection probabilities among track plots and During each of the surveys in April and November surveys (Jessop et al. 2013a). 2012 and 2013, we conducted 40 active diurnal The track-based index of goanna activity was searches (20 sites located near dams, 20 near tanks), derived by scoring the occurrence (presence/absence) during the survey in September 2014 we conducted 38 of tracks crossing 50 m track plots located along road searches (20 sites located near dams, 18 near tanks). transects radiating from dams and tanks. The transects Active search sites comprised 1 ha (100 m 9 100 m) were situated on low-use single lane dirt roads plots and were spaced a minimum of 2 km apart and (Paltridge 2002; Read and Scoleri 2014). We surveyed located along the same transects used to survey goanna 403 track plots over a total of 201.5 km (Online activity. At each site, active searches were conducted Resource 2). Each track plot consisted of a 50 m road simultaneously by two observers who portioned their section that was cleared of tracks on the day before the search effort so that each observer restricted their 123 Invasive cane toads might initiate cascades 1839 Statistical analyses search to a 50 9 100 m quadrat within each site. We recorded reptiles encountered on the ground, under Cane toad abundance at AWP and along road logs, in litter, in grass and on stems and branches of trees. An observers’ experience bias was avoided by transects using random observer combinations for each survey. Each site was actively searched for 10 min (5 min per We analyzed differences in cane toad density in the direct vicinity of AWP using a generalized linear observer) and was conducted between 9:00 and 10:30 am. To prevent double counting, observers mixed model (GLMM) with a Poisson distribution and avoided walking the same paths twice. Sighted reptiles a log link function. To account for multiple sampling were identiﬁed to family level by their pattern and size between years, AWP identity was included as random and the microhabitat they were encountered in. All factor. We analyzed differences in cane toad density along 12 km road transects radiating from the two encountered skink species belonged to four genera (Carlia, Ctenotus, Lialis and Menetia), all encoun- different reservoir types using a generalized additive model (GAM) with a Poisson distribution and a log tered dragon species belonged to two genera (Amphi- bolurus and Diporiphora). The total number of link function. To account for the nested structure of the data, we included the identity of each transect and the individuals recorded during 10 min of active search was used as an index of the abundance of skinks and year in which the survey was conducted in as random factors. All GLMM and GAM analyses were per- dragons at each active search site. formed in R Version 3.0.3 using the ‘glmm’ and Mammal activity ‘mgcv’ libraries. To investigate the alternative hypotheses that habitat Goanna activity and small lizard abundance along transects disturbance by cattle or predation by dingoes or feral cats were factors inﬂuencing the abundance of goan- We combined the track and foraging pit indices of nas and/or smaller lizards, we recorded the presence of tracks of cattle, dingoes, and feral cats at each tracking goanna activity in our analysis and deﬁned a plot as indicating recent goanna activity if goanna tracks and/ plot. An index of cattle activity for each track plot and each active search site was expressed as the percentage or recent foraging pits were present. Because our road of track plots with fresh tracks within a 1.5 km radius. surveys revealed a decline of cane toad abundance at To account for the wide-ranging habitat of dingoes, distances of up to 3 km from dams, followed by a dingo activity was expressed as mean values obtained steady count to distances up to 12 km (see ‘‘Results’’ for each sub-site. Cat activity was omitted from the section), we divided our 12 km transects into two analyses owing to the low activity of cats (only 3.8% sections (i.e.\ 3 and[ 3 km). For the initial analysis of goanna activity and small lizard abundance along of the track plots contained cat tracks). transects we compared the averaged goanna activity and lizard abundance of each transect section between Fire history individual transects using GLMM with a normal distribution and log link function. Type of nearest The reduction of vegetation coverage by ﬁre is known to inﬂuence the abundance of sand goannas and AWP, distance to AWP and the interaction of type and distance were included as ﬁxed factors in the models. smaller lizard species (Letnic et al. 2004; Bird et al. 2014). To investigate whether differences in the ﬁre To account for the nested structure of the data, we history could explain differences in the abundance of included the identity of each transect and the year in goannas, skinks and dragons we obtained data on the which the survey was conducted in as random factors. ﬁre history of each active search site from the North Because our expectation in this study system was that even though the biomass of lizards (i.e., as small Australian Fire Information. ectotherms) would be relatively high, the distribution of individuals is expected to be very patchy. This reﬂects well known observations, that semi-arid lizards, as consequences of sensitivity to heterogeneity 123 1840 B. Feit et al. in structural habitat resources (e.g. ground vegetation the active search site), cattle activity (percentage of cover, course woody debris) and relatively small home plots with cattle tracks within 1.5 km of the active ranges, can be extremely variable in spatial occurrence search site) and the number of months since the last (Letnic et al. 2004). Our survey design thus considered ﬁre as well as with mean values obtained for each sub- that an increased number of plots sampled once, rather site for dingo activity (percentage of plots with dingo than fewer plots sampled repeatedly, would permit tracks) to account for the wide-ranging habitat of better encounter rates and less variation in lizard dingoes. Because the impact of cane toads on goannas detection in an otherwise very large study area. We was negatively correlated with increasing distance however acknowledge, that as a potential trade-off of from dams whereas increasing distance from tanks this approach, our measurements of naive count data was not correlated with goanna abundance (see could not account for imperfect detection (Guillera- ‘‘Results’’ section), we used the distance from the Arroita et al. 2014). Ideally, if time had permitted, we nearest dam to each of the active search sites as a would have performed repeated count surveys on a proxy for the impact of cane toads on goannas. We large number of sites to allow use of potentially more used a backwards step-wise elimination process for robust count estimation methods (Royle and Nichols model simpliﬁcation whereby the most non-signiﬁcant 2003). predictor variables were sequentially deleted until all interaction were signiﬁcant. The most parsimonious model was then selected using Akaike’s Information Structural equation modelling Criterion for small sample sizes (AICc) as that with the Because our initial analysis indicated signiﬁcant lowest AICc value (Burnham and Anderson 2002). differences in cane toad abundance, recent goanna Standardized path coefﬁcients were calculated by activity and small lizard abundance between transects normalizing data to fall within one standard deviation in the vicinity of tanks and dams (see ‘‘Results’’ of a mean centered on zero and the amount of variance section), we used piecewise SEM to further test explained by each ‘piece’ of the SEM (i.e., the goanna, hypotheses based on a priori knowledge of interactions skink and dragon models) was assessed using marginal hypothesized to occur between cane toads, goannas R values. The overall ﬁt of the SEM was assessed and smaller lizard species (Grace 2006). We con- using a Fisher C test and associated p value. The model structed our a priori SEM model based on trophic is a good representation of the data if the Fisher C p- cascade theory and prior knowledge of factors value is [ 0.05. All SEM analyses were performed in impacting the abundance of small terrestrial lizards. R Version 3.0.3 using the ‘piecewiseSEM’ library. As opposed to classical SEM, where covariance matrices are used, piecewise SEM uses localized Model justiﬁcation estimates to deduce direct and indirect effect pathways (Grace 2006; Colman et al. 2014). This approach Interaction pathways between variables were deter- allows the modelling of data that do not meet the mined by applying a priori knowledge, which resulted assumptions of classic SEM and the incorporation of in the following set of hypothesized pathways exogenous factors such as spatial dependence (Pasa- (Fig. 5a): (1) Cane toad activity should negatively nen-Mortensen et al. 2013; Colman et al. 2014). affect goannas owing to lethal ingestion; (2) the Localized estimates within the SEM were ﬁtted using foraging activity of goannas should negatively affect a GLMM (Poisson log-link function; skink and dragon the abundance of skinks and dragons (Olsson et al. models) or LMM (goanna model). To account for the 2005); (3) because of selective predation pressure, nested structure of the data we included the identity of goanna activity should have a stronger impact on each transect as a random factor. For the GLMM, an skinks than dragons (Sutherland 2011); (4) we used observation level random effect was included to distance to the nearest dam as a proxy for the impact of account for overdispersion (Harrison 2014). cane toads because the impact of cane toads on Our initial models were parameterized with values goannas was negatively correlated with distance from obtained for each active search site for the variables dams (see ‘‘Results’’ section); (5) dingo activity goanna activity (percentage of plots with recent should negatively affect goanna activity owing to goanna tracks and/or foraging pits within 1.5 km of predation (Paltridge 2002); (6) habitat modiﬁcation 123 Invasive cane toads might initiate cascades 1841 Goanna activity along transects resulting from grazing by livestock can have detri- mental effects on both the abundance of goannas and Overall, the probability of detecting recent goanna of smaller lizard species such as skinks and dragons (James 2003); (7) time since ﬁre was hypothesized to activity was 3.3 times greater at plots in the vicinity of tanks where toads were rare (0.50 ± 0.08) than dams positively affect populations of goannas (Bird et al. 2014) and to negatively affect skinks and dragons where toads were abundant (0.15 ± 0.03, F = 18.88, p \ 0.01). In the vicinity of dams, (Letnic et al. 2004). 1,399 recent goanna activity was 2.0 times greater within the [ 3 km transect sections (0.10 ± 0.04) than\ 3km Results transect sections (0.20 ± 0.03, F = 4.43, 1,399 p \ 0.05; Fig. 4a). Cane toad abundance at AWP and along road transects Small lizard abundance along transects Cane toads were 5.2 times more abundant in the direct We encountered 6.0 times more skinks during 10-min active searches along transects in the vicinity of dams vicinity of dams (mean ± 1SE = 15.0 ± 2.9) than tanks (2.4 ± 0.6; F = 32.68, p\ 0.001). For both where toads were abundant (0.13 ± 0.02) than tanks 1,39 where toads were rare (0.02 ± 0.01, F = 31.48, reservoir types, the number of cane toads encountered 1,198 along road transects during the wet season was p \ 0.001). In both the vicinity of dams and tanks, correlated negatively with distance from AWP skink abundance was greater within the \ 3km (Fig. 3). In the vicinity of dams we encountered transect sections (dams: 0.19 ± 0.03, tanks: relatively high numbers of cane toads within the ﬁrst 0.07 ± 0.02) than the[ 3 km transect sections (dams: 0.10 ± 0.01, tanks: 0.01 ± 0.004, F = 17.61, 3 km of transects in comparison to distances[ 3km 1,198 (Fig. 3). We did not encounter cane toads at distances p \ 0.01, Fig. 4b). We encountered 2.0 times more dragons during [ 3 km from tanks (Fig. 3). We found no spatial autocorrelation among the Pearson residuals generated 10-min active searches along transects in the vicinity of dams where toads were abundant (0.06 ± 0.01) by the model (Mantel test, r = 0.04, p = 0.2). than tanks where toads were rare (0.03 ± 0.001, F = 6.18, p \ 0.05). Dragon abundance did not 1,198 differ between the \ 3 and [ 3 km transect sections within transects from dams or tanks, respectively (F = 1.71, p = 0.19, Fig. 4c). 1,198 Structural equation modelling Our proxy for cane toad abundance and impact, that is distance from dams, had a strong positive correlation with goanna activity and a strong negative correlation with skink and dragon activity (Fig. 5b). Goanna activity was correlated negatively with skink activity (Fig. 5b). Thus, increasing distance from dams, because it had a positive correlation with goanna activity (path co-efﬁcient = 0.70), had a negative, Fig. 3 Average number of cane toads (Rhinella marina) indirect correlation with the abundance of skinks encountered on 500 m road sections along transects radiating (indirect path co-efﬁcient =- 0.62; 0.70; Fig. 5b). from AWP ﬁtted with two reservoir types, dams where toads were abundant (n = 4) and tanks where toads were rare (n = 5), Dingo activity was correlated negatively with goanna during a period of rainy conditions. Lines are regressions and activity (Fig. 5b) and cow abundance was correlated 95% conﬁdence limits ﬁtted by a generalized additive model positively with dragon activity (Fig. 5b). Marginal R with a Poisson distribution and log link function. Error bars values were relatively high for the skink and goanna indicate ± 1SE 123 1842 B. Feit et al. bFig. 4 a Average probability of encountering recent signs of sand goanna (Varanus gouldii) foraging activity within 50 m road sections (n = 403) and average number of skinks (b) and dragons (c) encountered during 10 min searches (n = 198) at distances of under 3 km and over 3 km from AWP ﬁtted with two reservoir types, dams where toads were abundant and tanks where toads were rare. Error bars indicate ± 1SE components of the SEM (0.48 and 0.35, respectively) but relatively low for the dragon component (0.04). The Fisher C p-value was [ 0.05 (Fisher C value = 17.32, p = 0.07), suggesting that our most parsimonious SEM was a good representation of the data. Discussion In accord with our a priori predictions, the results of this study demonstrate that: (1) cane toad abundance in the late wet season decreased with distance from AWP and their numbers were greater in the vicinity of dams than tanks when distance from AWP was held constant; (2) goanna activity increased with distance from dams and, when distance to water was set to mean distance, goanna activity was greater in the vicinity of tanks where toads were rare than dams where toads were abundant; (3) the abundance of skinks and dragons was greater in the vicinity of dams where toads were abundant than tanks where toads were rare and skink abundance of was negatively correlated with distance from AWP. In comparison to distance to dams (our proxy for the abundance and impact of cane toads), cattle activity, time since last ﬁre and mammalian predator activity were weak predictors of goanna activity and skink abundance. However, cattle activity was the best predictor of dragon abundance. Taken together, these ﬁndings provide support for the idea that the invasion of cane toads has propagated an ecological cascade whereby cane toad induced declines of goanna populations near toads’ dry season refuges have facilitated increased abundances of small lizards. More generally, our results provide support for the notion that terrestrial invaders may trigger ecological cascades that become evident as alternating spatial patterns in the abun- dances of species negatively and positively affected by invaders. 123 Invasive cane toads might initiate cascades 1843 2 2 Fig. 5 a A priori piecewise structural equation model (SEM) goannas, skinks and dragons. R = marginal R values; path describing the response of populations of sand goannas coefﬁcients (± 1 SE) are shown adjacent to arrows; dashed (Varanus gouldii), skinks (Scincidae) and dragons (Agamidae) arrows show negative interaction and solid arrows show positive to the cane toad invasion of semi-arid rangelands in northern interaction; ?= p value of 0.05–0.1, * = p value of 0.05–0.01, Australia. b Most parsimonious SEM explaining the activity of ** = p value of 0.01–0.001, *** = p value of \ 0.001 A short-coming of our study is that we did not rates of predation by goannas near dams has allowed experimentally manipulate cane toad abundance but for increased abundances of small lizards. This instead relied upon natural differences in cane toad hypothesis is supported by earlier studies showing activity resulting from the type of reservoir utilized at that cane toads are more abundant at dams than tanks AWP and distance from AWP (Letnic et al. 2014). (Feit et al. 2015), that goanna populations have Because natural systems are intrinsically variable, it declined following the invasion of cane toads (Grif- remains possible that confounding factors such as ﬁths and McKay 2007, Doody et al. 2009) and that differences in grazing pressure, vegetation type, suppression of goanna populations can drive increases geomorphology, ﬁre history and the activity of in the abundances of skinks and dragons (Olsson et al. mammalian predators may have contributed to the 2005; Doody et al. 2013; Read and Scoleri 2014). differences we observed (Underwood 1990). During Moreover, the stronger negative correlation between the design of our study, we attempted to control for goanna and skink activity as opposed to that between variation in these variables in two ways. First, we goannas and dragons is consistent with previous selected study sites with little variation in underlying studies showing that sand goannas consume skinks geology, vegetation type, land use and ﬁre history. more frequently than dragons (Losos and Greene Second, at each of our study sites we measured indices 1988; Sutherland 2011). Nevertheless, our results of cattle grazing activity, mammalian predator activity indicate that, in addition to goanna activity, factors not and time since last ﬁre, and included these variables as measured in this study could inﬂuence skink abun- alternative hypotheses to explain goanna activity and dance along road transects in our study system. The lizard abundances in our SEM. None of these variables strength of the indirect path coefﬁcient of the effect of explained as much variation in goanna activity or distance to dams on skink abundance mediated by skink abundance as the direct or indirect effects of our goanna activity is - 0.62 (i.e. 0.70 9- 0.88), proxy for cane toad abundance and impact, distance approximately half the strength of the direct pathway from dam. between dams on skinks (1.33), suggesting that Our results are consistent with the hypothesis that goannas explain about half of the response of skinks higher encounter rates between toads and goannas in to increasing distance to dams. This could reﬂect an the direct vicinity of dams was the driver of the underestimation of goanna foraging by the activity reduction in goanna activity near dams. In turn, lower indices used in this study or factors affecting skink 123 1844 B. Feit et al. activity not measured during our surveys. We there- and our capacity to observe accurately lizards during fore caution that controlled experiments are required count surveys. For example, if small lizards in high- to conﬁrm or refute our cane toad induced ecological goanna density areas are exposed to greater predation cascade hypothesis. risk and respond through adjustments in daily activity Central to our hypothesis that cane toads’ impacts patterns or microhabitat use, or variation in cryptic or on lizard assemblages decrease with distance from ﬂight initiation behavior this could affect our capacity dams is the idea that cane toads are largely restricted to to accurately count them (Vanhooydonck and Van refuge sites with water during the dry season and Damme 2003; Cooper and Wilson 2007; Cooper and disperse away from refuges during the wet season Frederick 2009). Whilst this is an important method- (Letnic et al. 2014). Thus, during the wet season, ological consideration, reduced lizard abundance due encounters between cane toads and goannas could to the respective contributions of direct mortality or occur anywhere in the landscape, while in the dry increased anti-predator behavior, nevertheless repre- season goannas would most likely encounter toads sents the two key processes through which predators within 500 m of permanent water (Florance et al. limit prey populations (Preisser et al. 2005). Further 2011). However, during the wet season, the density of study is needed to better understand how goanna cane toads and hence likelihood of a goanna encoun- predation inﬂuences small lizard abundance. Such tering a cane toad should decrease with distance from studies could include those that use a survey design that considers repeated lizard plot counts to enable refuges because as they disperse they are in effect diffusing away from a point source (Florance et al. abundance estimation analyses that consider imperfect 2011; Tingley et al. 2013). We contend then that the detection (e.g. Royal–Nichols abundance type occu- patterns in goanna and small lizard abundance we pancy models; Royal and Nichols 2003), or explicitly report are legacy effects that reﬂect higher encounter set out to detect non-consumptive predation inﬂuences rates between goannas near (\ 3 km) sources of (e.g. predator induced spatial avoidance using 2-spe- permanent water during both the wet season and dry cies occupancy models; Robinson et al. 2014). These season. studies alongside others that also directly quantify Our SEM analysis suggests that distance to the variation in small lizard anti-predator behavior across nearest dam, and thus the population density and spatial gradients of goanna predation risk would impact of cane toads, was not the only factor greatly aid understanding of toad induced trophic inﬂuencing the foraging activity of goannas and the consequences in this system. abundance of small lizards in our study. Goanna Direct population level impacts of cane toads have activity was also negatively correlated with the not been restricted to a single goanna species but a activity of dingoes and, contrary to our expectations, multitude of native predators in terrestrial and aquatic positively correlated with cattle activity. In contrast to ecosystems including other monitor lizards, croco- increasing distance to dams, however, the activity of diles, snakes and quolls (Shine 2010; Feit and Letnic cattle and dingoes had only weak effects on the 2015). Doody et al. (2006, 2009, 2013, 2015) demon- foraging activity of goannas in our SEM. Similarly, strated that cane toad-induced decline of goannas can habitat disturbance by cattle and the reduction of have cascading effects on species not predicted to be vegetation coverage by ﬁre had weak effects on lizard directly affected by cane toads such as small lizards, abundance in our SEM. Nevertheless, the correlation tree snakes, freshwater turtles and grain-eating birds in between the abundance of skinks and goanna activity riparian systems in northern Australia. Cane toads was stronger than the correlation between their have also been reported to have direct suppressive abundance and cattle activity or the time since last ﬁre. effects on their invertebrate prey, and to compete with We acknowledge that because our count estimates nesting birds for burrows (Boland 2004; Greenlees of lizard abundances did not consider imperfect et al. 2007; Feit et al. 2015). Given the extent of their detection, the large difference reported for lizard reported direct impacts, we contend that it is likely that abundances between tanks and dams may be overes- cane toads have also had indirect impacts on species timated. Such a consequence could arise if goanna which have strong interactions with species that have predation also induced increased lizard anti-predator declined following the toad invasion. Indeed, given behavior that could affect their detection probability the multitude of possible indirect interaction pathways 123 Invasive cane toads might initiate cascades 1845 Colman NJ, Gordon CE, Crowther MS, Letnic M (2014) Lethal they could potentially disrupt (Doody et al. control of an apex predator has unintended cascading 2009, 2013, 2015; Feit et al. 2015), we suspect that effects on forest mammal assemblages. Proc R Soc B the invasion of cane toads has affected a much greater 281:20133094. https://doi.org/10.1098/rspb.2013.3094 range of taxa than has thus far been reported. Cooper WE, Frederick WG (2009) Predator lethality, optimal escape behavior, and autotomy. Behav Ecol 21:91–96 Cooper WE, Wilson DS (2007) Beyond optimal escape theory: Acknowledgements Funding was provided by the Hermon microhabitats as well as predation risk affect escape and Slade Foundation and Mazda Foundation. We thank Frogwatch, refuge use by the phrynosomatid lizard Sceloporus virga- Parks & Wildlife NT and the managers of Dungowan and tus. Behaviour 144:1235–1254 Camﬁeld for their support. A. Feit provided valuable comments Corbin JD, D’Antonio CM (2004) Competition between native on earlier drafts of the manuscript. Animal census procedures perennial and exotic annual grasses: implications for an were conducted under the Northern Territory Parks and Wildlife historical invasion. Ecology 85:1273–1283. https://doi.org/ Commission permit 44073, approved by the Animal Care and 10.1890/02-0744 Ethics Committees of the University of New South Wales (12/ Croll DA, Maron JL, Estes JA, Danner EM, Byrd GV (2005) 103A) and Western Sydney University (A9776). Introduced predators transform subarctic islands from grassland to tundra. Science 307:1959–1961 Open Access This article is distributed under the terms of the Doody JS, Green B, Sims R, Rhind D, West P, Steer D (2006) Creative Commons Attribution 4.0 International License (http:// Indirect impacts of invasive cane toads (Bufo marinus) on creativecommons.org/licenses/by/4.0/), which permits unre- nest predation in pig-nosed turtles (Carettochelys stricted use, distribution, and reproduction in any medium, insculpta). Wildl Res 33:349–354 provided you give appropriate credit to the original Doody J, Green B, Rhind D et al (2009) Population-level author(s) and the source, provide a link to the Creative Com- declines in Australian predators caused by an invasive mons license, and indicate if changes were made. species. Anim Conserv 12:46–53. https://doi.org/10.1111/ j.1469-1795.2008.00219.x Doody J, Hall M, Rhind D, Green B, Dryden G (2012) Varanus panoptes (yellow-spotted monitor) diet. Herpetol Rev References 43:491–492 Doody J, Castellano C, Rhind D, Green B (2013) Indirect facilitation of a native mesopredator by an invasive spe- Anderson CB, Rosemond AD (2007) Ecosystem engineering by cies: are cane toads re-shaping tropical riparian commu- invasive exotic beavers reduces in-stream diversity and nities? Biol Invasions 15:559–568. https://doi.org/10.1007/ enhances ecosystem function in Cape Horn, Chile. s10530-012-0308-8 Oecologia 154:141–153. https://doi.org/10.1007/s00442- Doody J, Mayes P, Clulow S et al (2014) Impacts of the invasive 007-0757-4 cane toad on aquatic reptiles in a highly modiﬁed ecosys- Anson JR, Dickman CR, Handasyde K, Jessop TS (2014) tem: the importance of replicating impact studies. Biol Effects of multiple disturbance processes on arboreal ver- Invasions 15:2303–2309. https://doi.org/10.1007/s10530- tebrates in eastern Australia: implications for management. 014-0665-6 Ecography 37:357–366. https://doi.org/10.1111/j.1600- Doody J, Soanes R, Castellano CM et al (2015) Invasive toads 0587.2013.00340.x shift predator–prey densities in animal communities by Baxter CV, Fausch KD, Murakami M, Chapman PL (2004) Fish removing top predators. Ecology 96(2544–2554):15031 invasion restructures stream and forest food webs by 0115152000. https://doi.org/10.1890/14-1332.1 interrupting reciprocal prey subsidies. Ecology Estes JA, Terborgh J, Brashares JS et al (2011) Trophic down- 85:2656–2663. https://doi.org/10.1890/04-138 grading of planet Earth. Science 333:301–307 Bird RB, Taylor N, Codding BF, Bird D (2014) Niche con- Feit B, Letnic M (2015) Species level traits determine positive struction and dreaming logic: aboriginal patch mosaic and negative population impacts of invasive cane toads on burning and varanid lizards (Varanus gouldii) in Australia. native squamates. Biodivers Conserv 24:1017–1029. Proc R Soc B 280:20132297 https://doi.org/10.1007/s10531-014-0850-z Boland CR (2004) Introduced cane toads Bufo marinus are Feit B, Dempster T, Gibb H, Letnic M (2015) Invasive cane active nest predators and competitors of rainbow bee-eaters toads’ predatory impact on dung beetles is mediated by Merops ornatus: observational and experimental evidence. reservoir type at artiﬁcial water points. Ecosystems Biol Conserv 120:53–62. https://doi.org/10.1016/j.biocon. 18:826–838. https://doi.org/10.1007/s10021-015-9865-x 2004.01.025 Florance D, Webb JK, Dempster T et al (2011) Excluding access Brown GP, Phillips BL, Webb JK, Shine R (2006) Toad on the to invasion hubs can contain the spread of an invasive road: use of roads as dispersal corridors by cane toads (Bufo vertebrate. Proc R Soc B 278:2900–2908 marinus) at an invasion front in tropical Australia. Biol Gooden B, French K, Turner PJ (2009) Invasion and manage- Conserv 133:88–94. https://doi.org/10.1016/j.biocon. ment of a woody plant, Lantana camara L., alters vegeta- 2006.05.020 tion diversity within wet sclerophyll forest in southeastern Burnham K, Anderson D (2002) Model selection and multi- Australia. For Ecol Manag 257:960–967. https://doi.org/ model inference: a practical information-theoretic 10.1016/j.foreco.2008.10.040 approach. Springer, New York 123 1846 B. Feit et al. Grace JB (2006) Structural equation modeling and natural sys- herbs. Oecologia 139:359–375. https://doi.org/10.1007/ tems. Cambridge University Press, Cambridge s00442-004-1518-2 Green B, King D (1978) Home range and activity patterns of the Olsson M, Wapstra E, Swan G et al (2005) Effects of long-term sand goanna, Varanus gouldii (Reptilia: Varanidae). Wildl fox baiting on species composition and abundance in an Res 5:417–424 Australian lizard community. Austral Ecol 30:899–905. Greenlees MJ, Brown GP, Webb JK et al (2007) Do invasive https://doi.org/10.1111/j.1442-9993.2005.01534.x cane toads (Chaunus marinus) compete with Australian Paltridge R (2002) The diets of cats, foxes and dingoes in frogs (Cyclorana australis)? Austral Ecol 32:900–907. relation to prey availability in the Tanami Desert, Northern https://doi.org/10.1111/j.1442-9993.2007.01778.x Territory. Wildl Res 29:389–403 Grifﬁths A, McKay J (2007) Cane toads reduce the abundance Parker I, Simberloff D, Lonsdale W et al (1999) Impact: toward and site occupancy of Merten’s water monitor (Varanus a framework for understanding the ecological effects of mertensi). Wildl Res 34:609–615. https://doi.org/10.1071/ invaders. Biol Invasions 1:3–19 WR07024 Pasanen-Mortensen M, Pyyko ¨ nen M, Elmhagen B (2013) Guillera-Arroita G, Lahoz-Monfort JJ, MacKenzie DI, Wintle Where lynx prevail, foxes will fail—limitation of a BA, McCarthy MA (2014) Ignoring imperfect detection in mesopredator in Eurasia. Glob Ecol Biogeogr 22:868–877. biological surveys is dangerous: a response to ‘ﬁtting and https://doi.org/10.1111/geb.12051 interpreting occupancy models’. PLoS ONE 9:e99571 Preisser EL, Bolnick DI, Benard MF (2005) Scared to death? Harrison XA (2014) Using observation-level random effects to The effects of intimidation and consumption in predator– model overdispersion in count data in ecology and evolu- prey interactions. Ecology 86:501–509 tion. PeerJ 2:e616 Read J, Scoleri V (2014) Ecological implications of reptile James C (2003) Response of vertebrates to fenceline contrasts in mesopredator release in arid South Australia. J Herpetol grazing intensity in semi-arid woodlands of eastern Aus- 49:64–69 tralia. Austral Ecol 28:137–151 Risbey DA, Calver MC, Short J et al (2000) The impact of cats James CD, Landsberg J, Morton SR (1999) Provision of and foxes on the small vertebrate fauna of Heirisson Prong, watering points in the Australian arid zone: a review of Western Australia. II. A ﬁeld experiment. Wildl Res effects on biota. J Arid Environ 41:87–121. https://doi.org/ 27:223–235 10.1006/jare.1998.0467 Robinson QH, Bustos D, Roemer GW (2014) The application of Jessop T, Kearney M, Moore J et al (2013a) Evaluating and occupancy modeling to evaluate intraguild predation in a predicting risk to a large reptile (Varanus varius) from feral model carnivore system. Ecology 95:3112–3123 cat baiting protocols. Biol Invasions 15:1653–1663 Rodriguez LF (2006) Can invasive species facilitate native Jessop T, Letnic M, Webb JK, Dempster T (2013b) Adreno- species? Evidence of how, when, and why these impacts cortical stress responses inﬂuence an invasive vertebrate’s occur. Biol Invasions 8:927–939. https://doi.org/10.1007/ ﬁtness in an extreme environment. Proc R Soc B. https:// s10530-005-5103-3 doi.org/10.1098/rspb.2013.1444 Roemer GW, Donlan CJ, Courchamp F (2002) Golden eagles, Letnic M, Dickman C, Tischler M et al (2004) The responses of feral pigs, and insular carnivores: how exotic species turn small mammals and lizards to post-ﬁre succession and native predators into prey. Proc Natl Acad Sci USA rainfall in arid Australia. J Arid Environ 59:85–114. https:// 99:791–796. https://doi.org/10.1073/pnas.012422499 doi.org/10.1016/j.jaridenv.2004.01.014 Roy HE, Adriaens T, Isaac NJB et al (2012) Invasive alien Letnic M, Koch F, Gordon C et al (2009) Keystone effects of an predator causes rapid declines of native European lady- alien top-predator stem extinctions of native mammals. birds. Divers Distrib 18:717–725. https://doi.org/10.1111/ Proc R Soc B 276:3249–3256. https://doi.org/10.1098/ j.1472-4642.2012.00883.x rspb.2009.0574 Royal JA, Nichols JD (2003) Estimating abundance from Letnic M, Webb J, Jessop T et al (2014) Artiﬁcial water points repeated presence–absence data or point counts. Ecology facilitate the spread of an invasive vertebrate in arid Aus- 84:777–790 tralia. J Appl Ecol 51:795–803. https://doi.org/10.1111/ Royle JA, Nichols JD (2003) Estimating abundance from 1365-2664.12232 repeated presence–absence data or point counts. Ecology Letnic M, Webb JK, Jessop TS, Dempster T (2015) Restricting 84:777–790 access to invasion hubs enables sustained control of an Sax DF, Stachowicz JJ, Gaines SD (2005) Species invasions: invasive vertebrate. J Appl Ecol 52:341–347 insights into ecology, evolution and biogeography. Sinauer Losos JB, Greene HW (1988) Ecological and evolutionary Associates Incorporated, Sunderland implications of diet in monitor lizards. Biol J Linn Soc Shine R (2010) The ecological impact of invasive cane toads 35:379–407 (Bufo marinus) in Australia. Q Rev Biol 85:253–291 Lunney D, Barker J (1986) Survey of reptiles and amphibians of Simon KS, Townsend CR (2003) Impacts of freshwater invaders the coastal forests near Bega, NSW. Aust Zool 22:1–9. at different levels of ecological organisation, with https://doi.org/10.7882/AZ.1986.001 emphasis on salmonids ad ecosystem consequences. Mack RN, Simberloff D, Mark Lonsdale W et al (2000) Biotic Freshw Biol 48:982–994 invasions: causes, epidemiology, global consequences, and Strayer DL (2010) Alien species in fresh waters: ecological control. Ecol Appl 10:689–710 effects, interactions with other stressors, and prospects for Miller KE, Gorchov DL (2004) The invasive shrub, Lonicera the future. Freshw Biol 55:152–174. https://doi.org/10. maackii, reduces growth and fecundity of perennial forest 1111/j.1365-2427.2009.02380.x 123 Invasive cane toads might initiate cascades 1847 Sutherland DR (2011) Dietary niche overlap and size parti- Proc Natl Acad Sci U S A 111:10233–10238. https://doi. tioning in sympatric varanid lizards. Herpetologica org/10.1073/pnas.1405766111 67:146–153 Underwood AJ (1990) Experiments in ecology and manage- Terborgh J, Estes J (2010) Trophic cascades: predators, prey and ment: their logics, functions and interpretations. Aust J the changing dynamics of nature. Island Press, Washing- Ecol 15:365–389 ton, DC Vanhooydonck B, Van Damme R (2003) Relationships between Thoresen JJ, Towns D, Leuzinger S, Durrett M, Mulder CP, locomotor performance, microhabitat use and antipredator Wardle DA (2017) Invasive rodents have multiple indirect behaviour in lacertid lizards. Funct Ecol 17:160–169 effects on seabird island invertebrate food web structure. Vitousek P, D’Antonio C (1997) Introduced species: a signiﬁ- Ecol Appl 27:1190–1198 cant component of human-caused global change. N Z J Tingley R, Phillips BL, Letnic M, Brown GP, Shine R, Baird SJ Ecol 21:1–16 (2013) Identifying optimal barriers to halt the invasion of White EM, Wilson JC, Clarke AR (2006) Biotic indirect effects: cane toads Rhinella marina in arid Australia. J Appl Ecol a neglected concept in invasion biology. Divers Distrib 50:129–137 12:443–455. https://doi.org/10.1111/j.1366-9516.2006. Tingley R, Vallinoto M, Sequeira F, Kearney MR (2014) 00265.x Realized niche shift during a global biological invasion.
– Springer Journals
Published: Jan 19, 2018