Feral frogs, native newts, and chemical cues: identifying threats from and management opportunities for invasive African Clawed Frogs in Washington state

Introduction

Invasive species threaten biodiversity globally (Didham et al., 2005; Didham et al., 2007; Pyšek & Richardson, 2010; Ahmed et al., 2022). While some effects of invasive species on native species and ecosystems are easily recognizable, other effects are challenging to identify. In some cases, native species responses that are behaviorally mediated may not be easily measured (Simberloff, 2013). Understanding the impacts of invasive species on a particular species or ecosystem is essential for appropriately allocating resources and coordinating management efforts (Epanchin-Niell, Englin & Nalle, 2009).

Amphibians globally have experienced tremendous losses and an estimated 41% of amphibian species are listed as threatened on the International Union for Conservation of Nature Red List (IUCN, 2024). Invasive species have contributed greatly to these declines as roughly 16% of threatened amphibian declines and approximately 30% of amphibian extinctions are at least partially attributed to invasive species (Falaschi et al., 2020). The threat of invasive species to amphibians may be greatest from aquatic invasive predators (Kats & Ferrer, 2003) due to predation, competition, hybridization, and disease (Falaschi et al., 2020). In North America, for example, native amphibians are not only threatened by invasive species in general but by competition with and predation by invasive amphibians. For example, bullfrogs (Lithobates catesbeiana) are invasive in the Pacific Northwest and threaten common amphibians, like red-legged frog (Rana aurora), to imperiled species like the Oregon spotted frogs (Rana pretiosa; Meshaka et al., 2022). The overall threat amphibians face globally and the impact of invasive species on this taxa has, in some circumstances, given rise to conservation interventions.

Management approaches for aquatic invaders have been trialed to control the impact on native species. A range of management techniques are used, including trapping and removal or euthanization, habitat modification, and chemical poisoning (Adams & Pearl, 2007; Lorrain-Soligon et al., 2021; Ojala-Barbour et al., 2021). One technique used with varying success for a range of invasive species includes the use of biocontrols. Biocontrols are living organisms that are introduced to an area or whose populations are enhanced to reduce an invasive species’ population or impact (Stoner, 2023). While some biocontrol management plans have introduced new problems to ecosystems, the use of native biocontrols has been a successful approach in others (Messing & Wright, 2006). For example, large-bodied groupers (Epinephelus striatus and Myceteroperca tigris) have been found to actively consume invasive lionfish (Pterios spp.) in the Caribbean (Mumby, Harborne & Brumbaugh, 2011). As such, the act of helping to amplifying native species in some locations may bolster biocontrol efforts.

Animal behavior analyses have become essential tools for conservation and have aided in identifying the impacts of invasive species and effective management techniques (Holway & Suarez, 1999; Berger-Tal et al., 2011). Amphibians are a model species for understanding the role of chemical cues in mediating predator–prey relationships and various non-consumptive interactions (Kiesecker, Chivers & Blaustein, 1996; Grayson et al., 2012). For instance, when presented with a visual cue, western toad (Anaxyrus boreas) tadpoles did not exhibit antipredator behavior, however in the presence of a predator chemical cue they display avoidance behaviors (Kiesecker, Chivers & Blaustein, 1996). These same types of analyses can be informative for understanding invasive species impacts as well. For example, Pacific chorus frog (Pseudacris regilla) tadpoles exhibit avoidance behavior when exposed to chemical cues of invasive bullfrogs (L. catesbeiana; Chivers et al., 2001). Further, R. aurora tadpoles exhibited high anti-predator refuge use behavior in response to both native and invasive fish and crayfish predator chemical cues, whereas chorus frog (P. regilla) tadpoles only responded to native fish predators but not invasive fish or crayfish chemical cues (Pearl et al., 2003). R. aurora also showed an increase in antipredator behavior when introduced to chemical cues for metabolic waste of conspecific tadpoles, showing a reduction in movement as a main response (Kiesecker et al., 1999), yet their behavioral responses to introduced bullfrogs appeared to vary by population (Kiesecker & Blaustein, 1997). The ability to identify and apply anti-predator responses can provide native amphibian populations a critical survival advantage, however—and from the perspective of the invader—the ability for non-native populations of amphibian to be able to identify the chemical cues of native threats also would provide them a survival advantage. For example, studies have found that non-native amphibians, like L. catesbeiana and Coquí (Eleutherodactylus coqui), sometimes cannot recognize cues from native predators (Garcia et al., 2012; Marchetti & Beard, 2021). L. catesbeiana recognized fish predators (e.g., largemouth bass, Micropterus salmoides) only if they were from a population sympatric with the predator (Garcia et al., 2012). Taken together, and whether from the perspective of the native or non-native amphibian, a species’ ability to recognize and react to chemical cues from taxa which they do not share a recent evolutionary history with can have major impacts on their success. This is particularly true when the dynamics of species interactions change, such as the recent arrival or expansion of a non-native species takes place. African clawed frogs (Xenopus laevis) are a feral amphibian that has potential to have large impacts on native amphibians and as their invasive range increases, this warrants research to investigate native species’ behavioral response—as well as their own.

X. laevis are native to Southern Africa (Van Sittert & Measey, 2016), but have been introduced to many countries around the world (Measey et al., 2012). X. laevis prey voraciously on a diversity of invertebrate and vertebrate animals in freshwater ecosystems (Fibla et al., 2020; Lillo, Faraone & Lo Valvo, 2011) and are likely successful invaders due to their generalist diet (Courant et al., 2017) and fast maturation times (Rödder et al., 2017). As such, there is concern that invasive X. laevis may outcompete native species for shared prey items or directly consume and extirpate native species (Rödder et al., 2017). In the United States, X. laevis have become well established in Arizona, California, Florida, and Washington (Ojala-Barbour et al., 2021). X. laevis in Washington are particularly troublesome because they have spread across multiple cities and counties in the south Puget Sound area and the frogs seem to persist in ponds that freeze in winter (Ojala-Barbour et al., 2021). Although X. laevis were first discovered in Washington in 2015, the threat of X. laevis to native species in Washington or the broader Pacific Northwest region remains largely unknown, as well as the degree of its spread beyond the three known regions where it currently occurs. Determining the threat to native aquatic species could help identify and refine management targets (Ojala-Barbour et al., 2021). However, current management tools for X. laevis in Washington are also sparse as prior eradication efforts using trapping and poisoning have failed (Ojala-Barbour et al., 2021). Thus, there is an urgent need to understand how much of a threat X. laevis pose to native species, particularly in this region, and what tools might be available to manage X. laevis.

To address the knowledge gaps in our understanding of the degree of threat X. laevis pose to native amphibians we used chemical behavioral analyses to explore the threat of and management options for this feral frog. First, we tested whether larvae of a native amphibian species, R. aurora, respond to chemical cues from feral X. laevis differently than to native amphibian predator chemical cues. This goal emerged from observations showing that ponds without X. laevis have diverse native amphibian communities whereas adjacent ponds with X. laevis are devoid of native amphibian species (Fig. 1; Friesen et al., unpubl). Second, we assessed whether native rough-skinned newts (Taricha granulosa) could be an effective biocontrol against X. laevis by testing whether feral X. laevis responded to newt chemical cues (including toxins). This goal emerged from two observations in early 2022. First, students at Saint Martin’s University began assisting the migration of newts across fence barriers that were meant to stop X. laevis from spreading (Fig. 2). Although we were regularly catching, marking, and releasing X. laevis in the preceding fall, our trapping in Lacey, WA yielded no X. laevis once additional newts were added to the pond, despite concurrent trapping effort in Issaquah (similar latitude, ∼100 km east) that yielded hundreds of X. laevis in similar sized ponds over the same timeframe. Second, we temporarily housed an X. laevis with a T. granulosa in our husbandry facilities which resulted in the X. laevis dying in less than 24 h. These two observations led to the hypothesis that X. laevis avoided and/ or were harmed by T. granulosa toxins or other cutaneous chemicals. T. granulosa is native to Western Washington and with other members of the genus Taricha have been the subject of intense study due to their robust cutaneous toxins, particularly tetrodotoxin (TTX; Vaelli et al., 2020). Research has shown that aqueous toxins exuded from these newts can elicit an antipredator behavioral response in larval amphibians, reduce the predatory success of dragonfly larvae, and cause invasive snails to migrate away (Zimmer et al., 2006; Bucciarelli & Kats, 2015; Ota et al., 2018). Accordingly, we predicted that native amphibian larvae would elicit an anti-predator response to a native newt but not an X. laevis and that X. laevis would be deterred by T. granulosa chemical cues.

Methods

Predator cues

We housed both partial egg masses together and R. aurora embryos hatched in aged tap water at room temperature from 25 May–22 June 2022. We exposed tadpoles (Gosner stages 24–42) to chemical cues from T. granulosa (a native newt predator) and feral X. laevis. The tadpoles developed during the trials. R. aurora tadpoles were collected from an adjacent pond where no X. laevis were present. We made a chemical stimulus solution by soaking an adult newt or X. laevis in 300 mL of aged tap water for two hours in separate 0 .47 L containers (Fig. 3). Untreated aged tap water was used as a control. After 2 h, the adult amphibians were returned to the housing enclosures. We pipetted two mL of the X. laevis cue, newt cue, or control water into R. aurora tadpole experimental containers containing 200 mL of aged tap water. The tadpoles were allowed 2 min of acclimation prior to recording behaviors. After the acclimation period, we recorded tadpole behaviors for 10 min. At least three trials of each treatment were conducted each day. We completed 90 trials (28 X. laevis cues and 31 each for newt cues and controls) using a total of 17 R. aurora tadpoles. Over the duration of our study, we exposed most tadpoles to all three treatments (control and two cue treatments), although some tadpoles were only exposed to two different treatment types across the study due to logistical constraints. Three replicates of each treatment were done each day and tadpoles were assigned to treatments to ensure they were exposed to different treatments in subsequent trials. Experiments occurred at room temperature and no refugia were added given the small size of the experimental containers. We scored R. aurora larval behaviors into four behavior categories and recorded duration of each: Nothing, Foraging, Swimming, and Frantic Swimming. We defined “Nothing” as sedentary tadpoles displaying no movement, “Foraging” as tadpoles exhibiting mouth movements and pecking at the bottom of the experimental containers, “Swimming” as constant, slow movements in circular patterns around the containers, and “Frantic Swimming” as rapid, erratic movements in variable directions.

Native newt biocontrol

Between 9 June 2022 and 8 September 2022, we performed behavioral choice tests on X. laevis exposed to T. granulosa to test whether X. laevis responded to newt cues. Adult X. laevis and newts were used in the biocontrol experiment and each animal was randomly selected from our husbandry facility. Each choice test was conducted in 2 L of aged tap water inside of a rectangular 38 L aquarium. The aquarium was divided into five sections along the long axis (Fig. 4). Mesh pouches made of black window screening were placed inside of the aquarium, adjacent and parallel to each of the two short sides (Fig. 4). One pouch was empty (control) and the other contained a newt (treatment). At the initiation of the experiment, we manually agitated each newt for 1 min, by gently stroking the anterior and posterior sides to promote the production of tetrodotoxin (Bucciarelli & Kats, 2015). Newt movement and direct interaction were 180° constrained by the use of sealed pouches but still allowed X. laevis to be exposed to chemical and visual stimuli. An X. laevis was placed in the center of the tank, parallel to the mesh pouches and facing out of the aquarium. For 10 min post-release, we observed X. laevis behavior and the duration spent at various positions within the enclosure. We recorded X. laevis positions based on where they occurred across the five sections in the enclosure and the total amount of time spent in each section. When the X. laevis was on the section with the newt or the section adjacent to the newt, its position was recorded as “Newt” (Fig. 4). When the X. laevis were in the middle fifth section, the time was recorded as “Center”. When the X. laevis was on the section with the empty mesh bag or the section adjacent to the empty mesh bag, the X. laevis’s position was recorded as “Away” from the newt. We performed a total of 50 X. laevis behavioral choice tests: 25 with the newt on the southwest side of the aquarium and 25 with the newt on the northeast side of the aquarium. We switched which side of the aquarium that newts were placed to ensure X. laevis were not responding to other confounding cues in the laboratory.

Results

Predator cues: Our models on individual behaviors found differences in R. aurora behavior (p = 0.03). Tukey’s post-hoc tests found that R. aurora tadpole Swimming rates were reduced in the newt treatment compared to Control treatments (p = 0.05). Tadpole rates in response to X. laevis were statistically indistinguishable from both the Control (p = 0.87) and Newt treatments (p = 0.15). For Frantic Swimming (p = 0.09) and Foraging (p = 0.89), our models found no differences in R. aurora behavior among treatments.

Native biocontrol: Linear mixed effects models and likelihood ratio tests supported a model containing only the variable Choice (Fig. 5; p = 7.99 e -14). Tukey’s post hoc tests found that all pairwise comparisons were significant (Center vs Away p = 1.0 e −0.4, Newt vs Away p = 0.003, Newt vs Center p = 1.0 e -04) such that feral X. laevis spent the least time in the Center third of the tanks (mean = 8.9 s, ± 1.0 s SE), intermediate amounts of time Away from newts (mean = 210.6 s, ± 4.9 s SE), and the most time next to the newts (mean = 368.5 s, ± 5.1 s SE).

TTX Analysis: We did not detect TTX in the sample solutions. Chromatograms showed no peak at the standard-derived elution time for TTX (Figure A1). It is possible that there are TTX analogues in the sample, but without commercially available standards, identification in the scope of this study is not possible. In general, the lack of a TTX peaks in the chromatogram indicates that TTX was at concentrations lower than 10x-15 moles/liter or possibly not present.

Discussion

Our study adds to the growing body of evidence that feral X. laevis pose a threat to native aquatic species (Kruger et al., 2019; Lafferty & Page, 1997; Lillo, Faraone & Lo Valvo, 2011). Feral X. laevis may be a concerning, hard-to-manage invasive predator in the Pacific Northwest. Our results show that a native species may not recognize X. laevis as a predator and that toxic T. granulosa may be challenging to use as native biocontrols against X. laevis; at least in the short-term. In our experiments, native R. aurora tadpoles exhibited strong anti-predator responses to native newt chemical cues by decreasing Foraging and increasing Frantic Swimming, but did not respond to X. laevis chemical cues. Interestingly, despite native tadpoles responding strongly to newt chemical cues, feral X. laevis did not respond to newts. These results underscore the threats that X. laevis poses to native species as a predator with few effective management options (Ojala-Barbour et al., 2021).

R. aurora tadpoles exhibit more antipredator behavior towards native newts than to feral X. laevis. Newts elicited a classic anti-predator behavioral syndrome in tadpoles by causing tadpoles to be sedentary with bouts of Frantic Swimming compared to more typical cruising Swimming and Foraging behaviors (Watkins, 1996; Laurila, Kujasalo & Ranta, 1997; Van Buskirk & Mccollum, 2000; Bridges, 2002; Gabor et al., 2019). X. laevis cues elicited no such response in R. aurora tadpoles. These findings suggest that native Pacific Northwest amphibians have evolved to exhibit antipredator behavior towards native predators but are unable to recognize invasive amphibian predator cues. This indicates R. aurora tadpoles are potentially vulnerable to X. laevis predation. However, the overall predation risk from X. laevis to R. aurora remains unclear as we did not perform feeding trials. Further research could clarify whether invasive X. laevis consume native amphibian larvae at high enough rates to cause population-level impacts. Additionally, because continued exposure to a predator cue can change the response of the cue receiver, it is possible that responses could have changed over the course of the trials (Kruger et al., 2019).

For this work, we focused on antipredator behaviors in R. aurora tadpoles—a species that is regionally listed as Stable, despite experiencing population declines primarily due to forest loss (Washington Herp Atlas, 2009), but has been listed as imperial is other part of its range (e.g., within Canada; Environment Canada, 2016). Beyond the direct potential impacts to R. aurora, like predation, this work highlights how X. laevis may be a threat to other native species—including more sensitive species—which may not recognize it as a predator. For instance, our source X. laevis population in Lacey, WA is less than 35 km away from known populations of federally threatened R. pretiosa in Thurston County. Given the close proximity, of invasive X. laevis to federally listed amphibians, there is a need to proactively manage the spread of X. laevis and understand impacts to sensitive species, particularly if these species are naive to X. laevis predator cues. Beyond impacts to amphibians, there is also a need to understand potential impacts to native fishes. For instance, the same nearby habitats that host federally threatened R. pretiosa also are home to olympic mudminnows (Novumbra hubbsi), a state-sensitive species that is small (<80 mm long) and potentially vulnerable to X. laevis predation. Furthermore, diverse salmonid species occur near invasive X. laevis populations in Washington (Ojala-Barbour et al., 2021) that may also fall with this invasive frog’s dietary range. Salmonids in the Pacific Northwest are culturally, ecologically, and economically important and several are listed under the U.S. Endangered Species Act (Quinn, 2004). Invasive X. laevis have been repeatedly detected in and adjacent to water bodies with various salmon species, including kokanee salmon (Oncorhynchus nerka). Although adult salmon are too large for X. laevis to consume, embryonic and fry life stages may be vulnerable to X. laevis predation, particularly if salmon are naive to X. laevis predator cues.

We anticipated that native newts might serve as a potential biocontrol agent against X. laevis. Although the neurotoxin TTX has been extensively studied in Taricha newts for its anti-predatory properties (Zimmer et al., 2006; Bucciarelli & Kats, 2015; Ota et al., 2018), to our knowledge it has not been studied for potential biocontrol purposes. We were motivated to test whether T. granulosa might be an effective biocontrol because several casual observations suggested that X. laevis may be sensitive to newt toxins. In particular, we anticipated that T. granulosa would be so toxic as to elicit a relatively rapid behavioral response in X. laevis. However, the presence of newts in our study appeared to attract rather than deter X. laevis. There are multiple reasons for this. First, we conducted relatively short-duration trials to assess X. laevis behavior. Longer trials may reveal different patterns if aqueous TTX takes longer than 10 min to influence X. laevis physiology. Second, additional work may benefit from testing different densities of newts as higher doses of TTX may be needed to influence X. laevis. Third, our experiments did not allow X. laevis to directly interact with newts. Although we attempted to stimulate TTX in the T. granulosa, our experimental design limited interspecific interactions that could have produced ecologically relevant exposures. Regardless, the potential utility of T. granulosa as a biocontrol is probably greater through passive toxicity rather than through consumption. Other types of biocontrol could include large invertebrates, which X. laevis have been shown to exhibit antipredator behavior to (e.g., measured as a decrease in activity when exposed to a predatory beetle, Dytiscus dimidiatus, and crayfish, Procambarus clarkii; Kruger et al., 2019). Finally, X. laevis may be attracted to visual cues more so than chemical ones. One study found that removing X. laevis was most successful when traps were baited with conspecifics (Lorrain-Soligon et al., 2021); this result, in tandem with our findings, suggests that X. laevis may generally respond to visual cues like movement. Future studies may benefit from testing the response of X. laevis strictly to chemical cues. Although the ability to produce a powerful neurotoxin makes Taricha newts a tantalizing potential candidate for X. laevis biocontrol, additional research is needed to assess if this is a viable and ecologically neutral management option.

While our research indicated that invasive X. laevis are chemically cryptic predators that could pose a risk to native species and which are not readily deterred by newt chemical cues (including toxins), the chemical mechanisms underlying the relationships we explored warrants further attention. The newts used in our research were collected at our field site and kept in a laboratory setting for 1–2 months prior to our experiments. Because of the conflicting observations that motivated our experiment and our experimental findings, we analyzed aqueous newt extracts to determine if TTX was present and estimated concentrations. This analysis found no detectable TTX in the solutions which may have affected chemical cues between the newts and X. laevis in this study. Even so, this analysis found possible TTX analogues and/or relevant metabolites. While some research has indicated that TTX may increase in captive newts (Hanifin, Brodie & Brodie, 2002), other research shows lower TTX levels in newts compared to wild individuals (Gall et al., 2022). There is also evidence that TTX is linked to the newt microbiome, potentially indicating our captive setting did not allow for proper microbe growth (Vaelli et al., 2020; Gall et al., 2022). Further, TTX concentrations vary and fluctuate within and among T. granulosa populations (Bucciarelli et al., 2016; Reimche et al., 2020) and so our population may inherently maintain low amounts of TTX or at the time of sampling possessed relatively low toxin concentrations. Interestingly, our results clearly show that native R. aurora tadpoles respond to newt chemical cues, regardless of whether TTX or some other possible analogue was the constituent molecule of the solution. It may have also suggested their ability to detect it in concentrations, while our methods could not. These findings highlight new opportunities for understanding the chemical ecology of newts and their interactions with other species.

Conclusion

We aimed to identify the roles that chemical cues play in mediating the relationships between invasive X. laevis and native amphibian prey and toxic newts. We found that: (1) native R. aurora tadpoles show strong anti-predator responses to newts but do not recognize X. laevis as predators and (2) X. laevis were attracted rather than deterred by T. granulosa chemical cues in short-duration trials. The lack of anti-predator responses to invasive X. laevis may provide a foraging advantage over native amphibian predators and suggest X. laevis have potential to have detrimental effects on native species populations. It is also possible that introduced X. laevis do not have a response to the defenses of native species because they have not co-evolved with the mechanism. Our work has begun to uncover some of the mechanisms that may allow X. laevis to threaten native species and highlights new areas of research to improve management of this global invader.

Supplemental Information