A taxonomically and geographically constrained information base limits non-native reptile and amphibian risk assessment: a systematic review

Representation of herpetofaunal groups in the literature

Only 1,116 species (6% of c. 18,145 described herpetofaunal species; Frost, 2017; Uetz, Freed & Hošek, 2017) are included in the 836 studies relating to herpetofaunal invasions in this review. A total of 552 (49%) of these species appear in only one study, and 909 species (81%) appear in fewer than five studies. A total of 95% of species were included in fewer than 12 studies. Hardly any studies were conducted on non-native herpetofauna before 1990, and most studies were published after 2000 (Fig. S1). A large proportion of papers (653/836, 78%) focused on a single species, and only a minority (54 or 6%) reported on more than 10 species.

Of the papers included in this review, most focus on or include frogs (58%, Table 1), while lizards appear in almost a quarter of studies. The remaining herpetofaunal groups are represented in 18% or fewer of papers relating to non-native reptiles and amphibians. However, frogs and lizards are the largest herpetofaunal groups and as such, the number of species from these groups present in the literature actually under-represents these groups as a whole (Fig. 2A). Turtles and crocodiles are comparatively small herpetofaunal groups (25 crocodile species and 347 testudine species have been described to date) and have the most representative sample of species covered by the literature. Both groups are overrepresented in the papers that were reviewed (Fig. 2A, 47% and 56% of species from these respective groups occur in at least one of the papers included in this review, Table 2), as are snakes.

Some families have received more attention than others. For reptiles, nearly all the testudine and crocodilian families (e.g. Cheloniidae, Testudinidae, Emydidae, Alligatoridae, Crocodylidae) are overrepresented in the literature, that is they have more species written about than expected by chance given the size of the family, compared to reptiles as a whole (Fig. 2B). Snake families that stand out include boids (Boidae and Pythonidae), Viperidae and Elapidae, while the lizard families most overrepresented across studies are iguanids (Iguanidae and Corytophanidae), varanids and chameleons. Over-represented amphibian families include three of the nine salamander families (Ambystomatidae, Cryptobranchidae and Salamandridae), and nine out of 56 frog families, including Ranidae, Pipidae, Dendrobatidae, Ceratophryidae and Bombinatoridae (see Fig. 2C for these and others). Families might be under-represented in the literature if (1) species in these families are being translocated without being reported, or if (2) species in these families are genuinely not moved around, typically resulting in no or very few non-native representatives for the group. The fleshbelly frogs (Craugastoridae, approximately 800 species) were the largest amphibian family with no species present in the invasion ecology literature. The largest reptile family with no representatives was the Uropeltidae (shieldtail snakes, 54 species).

There are a handful of well-studied species within each herpetofaunal group (Fig. 3). These include the cane toad (Rhinella marina; 243 papers, or 29% of all papers focus on or include this species), the American bullfrog (Lithobates catesbeianus; 130 papers), the red-eared slider (Trachemys scripta; 95 papers), the brown tree-snake (Boiga irregularis; 57 papers), the African clawed frog (Xenopus laevis; 51 papers) and the brown anole (Anolis sagrei; 41 papers, Figs. 3 and 4). The best-studied salamander is the tiger salamander (Ambystoma tigrinum, 15 papers), and the best-studied crocodilian is the spectacled caiman (Caiman crocodilus, 14 papers, Figs. 3 and 4). The literature on crocodiles, frogs and turtles is particularly skewed towards individual species. Over half (59%) of papers that include crocodiles focus on or include Caiman crocodilus, and half (50%) of the literature on frogs is focussed on or includes the cane toad R. marina (while L. catesbeianus appears in 27% of frog papers). The red-eared slider features in nearly three-quarters (73%) of literature on non-native turtles (Fig. 4). The literature on lizards is least dominated by a single taxon, with the most well-represented species, Anolis sagrei, appearing in 21% of papers that cover non-native lizards (Fig. 4).

Subject focus of research

Most research (416 papers; 50%) has been conducted on impacts (Fig. 4). A large portion of this impact literature (42%), however, covers impacts of only two species, R. marina and T. scripta (Fig. 4). Similarly, the literature on control of non-native herpetofauna is heavily biased in favour of cane toads (included in 33% of papers on control) and brown tree-snakes (24% of papers on control, Fig. 4), while cane toads appear in over a third of studies (39%) on invasion correlates. There was no significant difference in the distribution of literature across subjects between papers that focussed on one species and papers that covered multiple species or which made no specific mention of species (V = 21, p = 0.7422).

Geographic distribution of work

Excluding the 22 global studies, which largely made use of the same information base as the well-studied areas, the bulk of research on non-native herpetofauna covered in our review has been conducted in Australia (217 studies), the US (195 studies, mostly focussed on or including Florida, 86 studies and California, 41 studies), Brazil (40 studies) and Spain (40 studies), as well as several islands or island groups such as Guam (47 studies), Hawaii (46 studies) and the greater Caribbean (47 studies), with very limited information from other localities (Fig. 5). For example, the seven studies that were identified specifically from continental Africa, were all conducted in South Africa.

We further compared the geographical distribution and frequency of literature on herpetofauna to the distribution of documented introductions from Kraus (2009). One-third (33%) of the ∼600 species (nomenclature updated to match our list) included in the Kraus (2009) database have been studied in fewer localities than they have been introduced (Table 3). Excluding non-location-specific studies, only eight countries, five oceanic islands/island states and six US states have two or more location-specific studies per successful invasive amphibian or reptile introduction. The majority of localities (79%) where established populations have been recorded (n = 191: 145 countries and 46 US states) have fewer than two location-specific studies (aside from global or continental reviews) for every successful species (mean = 1.1), providing a poor basis for their risk analyses. Only six of the 23 localities that have more than 10 established non-native species, have more than two location-specific studies per successful species: Australia, Guam, the UK, China (inclusive of studies in Taiwan), California and France.

Additional anomalies are apparent when considering the number of species covered in the literature per location (n = 767 studies, Fig. 5B). In South Africa, where only seven established species have been recorded (Kraus, 2009), 285 species are covered by the local literature (seven studies, although the bulk of the species appear in a list of traded species), equating to 41 species documented per successfully established non-native. For localities with more than 10 established non-natives (n = 23), rates of study range from <0.08 species studied per successful species for La Reunion, to >3.5 species studied per successful species in Texas, the UK, Indonesia, Florida, China, Japan and the US as a whole. The US has the highest recorded number of successfully established non-natives (108 species), covered at a rate of four species studied per invasive species (451 species appear in the 192 papers that have a US focus). In the case of China (25 studies, including Taiwan, that document 47 species, at 4.3 species studied per established species), the number of successfully established species (11) is likely an underestimate (Liu, McGarrity & Li, 2012).

At a location level, there is clearly also selection towards studying certain species in particular places. For example, in Australia, where 28 established non-natives have been recorded (Kraus, 2009), 91% of studies focus on or include the cane toad. Similarly, 92% of the papers that include Guam and the Mariana Islands dealt with or include the brown tree snake, and 44% of papers that include China focus on or include the American bullfrog. Of the five most widely introduced species (Kraus, 2009: T. scripta, 84 localities (countries/US states), the Brahminy blind snake Indotyphlops braminus, 65 localities, L. catesbeianus, 58 localities, R. marina, 48 localities and the common house gecko Hemidactylus frenatus, 45 localities), only the American bullfrog—studied in 27 countries and 18 US states—has been studied in at least three quarters of the number of localities to which it has been introduced. The common house gecko (33%) and Brahminy blind snake (28%) have been studied in a third or less of the number of introduction locales.

Success of different herpetofaunal groups

According to the data contained in Kraus (2009), lizards and frogs have had the highest rates of successful establishment per species introduced (over 55% of frogs and lizards introduced outside their native range have established in at least one location), while crocodiles (at 14%) have the lowest (Table 2). However, crocodiles and turtles (∼27% of known species in both groups) have had the most representative sample of species introduced from their respective herpetofaunal groups (Fig. 2). It is therefore not surprising that turtles have the most representative sample of established or invasive species: 11.5% of all described turtles have successfully established somewhere outside their native range (Table 2). The proportion of successful non-native species is much lower for other groups (typically 1–2% and as low as 0.8% for snakes, Table 2).

Despite the high rates of establishment for those lizards and frogs that have been introduced, the low representation of species that have been introduced outside of their native ranges (according to the literature on which Kraus, 2009 is based) means that we know nothing about the invasive potential of the ∼6,200 lizards and ∼6,600 frogs that have never been given the opportunity to demonstrate their potential to establish or invade (Table 2).

Taxonomic biases

While few non-native herpetofaunal taxa have been studied in comparison to species from other groups (Dawson et al., 2017), the bias in studies on non-native herpetofauna is similar to the bias in information on native species. Less than half (40%) of reptile species have had their conservation status assessed (Bland & Böhm, 2016; Meiri & Chapple, 2016). Those families with the fewest conservation assessments included either families with no species identified in this review, or families that were under-represented (e.g. Amphisbaenidae). With the exception of Opluridae (Madagascan iguanas), from which no species were identified in our review, those families with the most conservation assessments (Meiri & Chapple, 2016) were identified to be either over-represented (Iguanidae) or proportionately represented (neither over nor underrepresented, e.g. Crotaphytidae, Phrynosomatidae) in the invasion literature. Amphibians, on the other hand, have all been assessed through The Global Amphibian Assessment (Stuart et al., 2004). All families identified through the global assessment to be threatened by over-exploitation (for at least one species in the family, Stuart et al., 2004) were also found to be well- or over-represented in the invasion literature (e.g. Ranidae, Dendrobatidae, Microhylidae, Ambystomatidae, Salamandridae, Cryptobranchidae, Fig. 2), highlighting that families or species that are used by people are more likely to become invasive.

For plants, grasses are an example of a huge family with many known invasive species. Yet only a tiny portion have been assessed to determine the extent of introduction to new localities and the level of establishment and invasion (Visser et al., 2016). Focussing on functional (Canavan et al., in press) or taxonomic groups (Canavan et al., 2017) has allowed scientists to distinguish syndromes of traits that enhance invasiveness. For herpetofauna, we suggest that a useful approach is to classify relevant functional groups of invaders and identify traits within these groups that correlate with invasive success (Allen et al., 2017; Tingley et al., 2010). For example, there is likely a difference in invasion correlates for species that are intentionally introduced (e.g. via the pet trade) and those that tend to move around as ‘hitch-hiker’ species (Kraus, 2007). Some groups, despite their large size, are absent from the invasion literature because no species have been introduced to localities outside their native range. For example, families that do not make good pets (e.g. snakes from the family Uropeltidae, Fig. 2) and/or are unlikely to be transported accidentally (e.g. groups that have small, remote distributions and very specialised habitat). Other families appear to be over-represented in the invasion literature. For example, nearly all testudine families are overrepresented. This makes sense because tortoises and turtles are popular pets and are therefore widely traded and documented outside of their native ranges. The dominance of the red-eared slider in the literature, however, suggests that although testudines in general are exceptionally widely traded, proportionately fewer species actually become invasive or have notable impacts. Red-eared sliders are traded in vast numbers (common estimates are upwards of 3 million hatchlings traded globally on an annual basis, Ramsay et al., 2007). Reducing the volume of this trade could be achieved by enforcing stricter controls on animal husbandry and trade, and encouraging the public to be more responsible pet keepers (Williams, 1999).

Geographic biases

For the large majority of countries, especially in Africa, where little information is available, there are few known established non-native species and few studies. There are, however, also countries where despite several (e.g. Egypt, Greece) or many (e.g. Japan, Indonesia) established species, very few studies have been conducted (Fig. 5C). For Japan and Indonesia, at least, the studies that have been conducted have covered a broader range of species than are known to be established (Fig. 5B). Peer-reviewed studies from these two countries, along with others like the UK, New Zealand and South Africa have covered many more species than are currently established (Fig 5B). The reason for this is not always clear, but could be due to a keen interest in herpetology in these areas or proactive research into trade and/or risk assessment (Goka, Okabe & Takano, 2013; Chapple et al., 2016).

In other locations, such as Australia and Guam, the peer-reviewed literature generation has been prolific (Fig. 5A), but focussed on single species. This means that although Australia has the most peer-reviewed literature (Fig. 5A) and the most papers produced per invasive species (Fig. 5C), most of the species that are established in Australia are still understudied (Fig. 5B). With the exception of a few places such as Florida where there is significant interest in the high numbers of invasions that appear to be driven by a large number of ports, a well-developed trade in exotic pets, and high levels of environmental disturbance (Krysko et al., 2016); the UK, which has a large number of English-language academic hubs; and Brazil, rates of study are worse for the rest of the globe. This likely reflects a lack of research capacity where there are invasions that are not reported or studied. For example, there are almost no studies or invasions recorded for Africa. This could be the result of fewer introductions and/or a capacity gap in herpetologists to report invasions. In other instances (e.g. China and Russia), it is possible ;that both invasions and literature on them have gone undetected because documentation regarding these events are not available in English, and therefore not searchable using the Web of Science (Adam, 2002). Records of first introductions for reptiles show exponential growth since about the 1950s, with no end in sight to this trend (Seebens et al., 2017). While there are likely already many undetected invasions, new areas remain vulnerable and increased awareness is important for preventing future invasions.

Setting priorities

Countries like New Zealand that have focussed on allowing importation and trade in only a short list of permitted species appear to have achieved the highest success in reducing introductions (Genovesi et al., 2015; Seebens et al., 2017). The current trend for assessing which species should be permitted on such lists is to quantify the impacts of the species based on published information. This begs the question: How many publications are needed to provide adequate information to allow for the accurate assessment of risk of invasiveness and impact? This is not an easily quantifiable number, but we can gain some insights using recent EICAT assessments and their confidence levels to suggest numbers. Without directed research, only 265 of 365 papers on 39 invasive amphibians had impacts that could be scored using the EICAT scheme (Kumschick et al., 2017). Of these, only eight species could be rated with high confidence for at least one impact mechanism and no species was rated with high confidence on more than one impact mechanism (EICAT scores 12 impacts, of which eight are applicable to reptiles and amphibians: Predation, Poisoning/Toxicity, Competition, Hybridisation, Disease Transmission, Interaction with other species, Parasitism, and Grazing/Herbivory/Browsing). Given that these eight species include the five most commonly studied frogs and salamanders (see Figs. 2B and 2D), we might conclude that even hundreds of studies are not sufficient to produce high-confidence scores on each impact mechanism for EICAT (ignoring SEICAT impacts). However, one study is capable of producing high confidence on one impact mechanism. Two such examples of species that were rated highly on confidence with just a single study: Dubey, Leuenberger & Perrin (2014) on hybridisation in Italian water frogs (Pelophylax bergeri) and Holsbeek et al. (2010) on hybridisation with Levant water frogs (P. bedriagae). This suggests that with more work focussed on such impact assessments, total impact for EICAT may be assessed with a minimum of eight papers, relating to each of the impact mechanisms in Hawkins et al. (2015, see above). To date, no EICAT assessment has found a comprehensive literature for any invasive species (Evans, Kumschick & Blackburn, 2016). From our work, we know that a minority of invasive herpetofaunal species have been the focus of any research. Clearly, much more directed work is needed.

Our study revealed that the majority of publications (78%) focus on single species, and that a large proportion of these, and other studies in this review (50%) concern impacts, which is good news for those hoping to score EICAT and SEICAT for these species. However, we caution that risk assessments require a fuller understanding of the invasive species, and that studies on pathways (11%) and trade (2%) are particularly poorly represented (Fig. 3). Given the importance of the first in risk assessment and the volume and key role of the latter in introductions (García-Díaz et al., 2017), we emphasise the need for more strategic publications analysing the trade in herpetofauna and other pathways related to their unintentional movement (Tingley et al., 2018).

The Global Amphibian Assessment is a good example of a world-wide initiative that drove substantial work to collate information for all species to provide a baseline and a fantastic resource for refining data. Given that only a small portion of herpetofauna currently appear to be moved around in high numbers, or show invasive tendencies, setting up more global initiatives to focus on groups of invaders across a broader geographic range is not an unrealistic task, particularly considering that distribution information is already available for amphibians. Initiatives for invaded countries to work together on the impacts of common invasive taxa could provide an important platform for accumulating crucial information on impacts. One such initiative in Europe saw members from four EU countries funded to work jointly on the impacts of the invasive African clawed frog, X. laevis: INVAXEN (http://www.anthonyherrel.fr/INVAXEN/). This initiative has added 15 published articles on this species, nearly doubling the available data on their invasive populations (Courant et al., 2017; Louppe, Courant & Herrel, 2017; Rödder et al., 2017). Funded by BIODIVERSA, this call did not include funding for studies on populations in non-participating EU countries (e.g. Italy), or in the native range of the species in southern Africa. There is scope for similar work on species like the red-eared slider, and other turtles that are currently studied in fewer locations than which they have been introduced (e.g. European pond turtles, Emys orbicularis and common snapping turtles Chelydra serpentina), several widespread gecko species, agamids like oriental garden lizards (Calotes versicolor), the Asian toad, and selected species from the families that are overrepresented in the literature and trade or even the cane toad outside of Australia.

Local and regional herpetological societies have a crucial role to play in this regard and should be encouraged to publish all new records of reptiles and amphibians in online databases and society websites. Many societies already do record such information in newsletters, but digitizing these data and making them available online could go a long way to improving the geographic coverage of literature and even reducing the taxonomic bias in published information. McGeoch et al. (2016) provide a protocol for prioritizing species, pathways and sites to assist countries in meeting Aichi Biodiversity Targets (Convention on Biological Diversity). Herpetological societies should contribute relevant information to the Global Register of Introduced and Invasive Species currently under development within the Global Invasive Alien Species Partnership framework (McGeoch et al., 2016). In the absence of information, risk assessments will continue to rely on information from models based on well-studied species. Improving the geographic coverage of studies on model organisms and then the taxonomic coverage of model taxa will go a long way to improving predictions for invasive species and ultimately reducing their impacts.