To generate our dataset, we first formalized a working definition of the term ‘non-native’ in the context of our research objectives. We defined non-native terrestrial gastropods as any species that has been either intentionally or accidentally introduced into CONUS and that is indigenous to areas outside of North America. As the geographic distributions of terrestrial gastropod species are generally understudied, native ranges of species documented only outside of CONUS but within North America might indeed include CONUS. Therefore, species that are native to Canada and Mexico are not considered in this study, nor are extralimital species that are native to portions of the U.S. but have been translocated to other regions within the country (e.g., Euglandina rosea Férussac, 1821). To identify non-native species’ records, we compiled all available information from state and federal governmental technical reports, scientific literature (e.g., Robinson 1999, Cowie et al. 2009), online data aggregators (GBIF, BISON), citizen science databases (iNaturalist), online collection portals specific to museum collections (InvertEBase), online data repositories specific to invasive or pest species (e.g., USDA APHIS, found at www.aphis.usda.gov/aphis/home/), and direct conversation with several malacologists (Robert Cowie, Daniel Dourson, Gerald Dinkins personal communication). Though we are confident in the overall breadth of our search efforts, we also acknowledge that this may represent an incomplete dataset.

Importantly, we also recognize the body of analytical and statistical quandaries associated with data sourced from citizen science networks and other large data aggregators (Bird et al. 2014; Kosmala et al. 2016; Bayraktarov et al. 2019). Our goal was to catalog and outline the potential pervasiveness of these non-native species by synthesizing all available data. Therefore, we note that we did not correct for spatial biases (e.g., spatial autocorrelation), nor did we adjust our data based on possible pseudo-replication, detection rates, or other common sources of geospatial data error. Thus, we encourage readers to treat our results as a synthesized dataset from which they can then begin to adjust for spatial biases for future geospatial modeling (e.g., species distribution modeling).

We utilized the online portal MolluscaBase (available at www.molluscabase.org) to verify the taxonomic identity of all species and to avoid double counting synonymous records. In cases of species being known by several taxonomic identities, searches for each identity were subsequently searched for, placed under the most updated synonym, and records were thoroughly searched by all authors to avoid overlap. When records were identified as erroneous, questionable, or of limited utility (e.g., falling outside CONUS or directly within the centroid of a county), they were removed from the dataset. If a detailed location description was provided for a record that did not contain a georeference, we georeferenced these records using the web application GEOLocate (available at www.geo-locate.org/).

Data were separated into three different sets for reporting: 1.) all records with or without georeferences, 2.) all records with georeferences, and 3.) all records with georeferences and temporal data. The second dataset with all georeferenced records was used for all downstream summaries beyond explicit analyses of spatiotemporal trends, for which the third dataset was used. Lastly, records were categorized by source: 1.) museum and natural history collections, 2.) state or federal governmental agency, 3.) scientific literature that did not already have records associated with a museum collection, and 4.) citizen science database.

Literature and geospatial data pertaining to each species identified as non-native in CONUS were reviewed and used to assign a continental origin with respect to the species’ indigenous range. Several species (e.g., Cornu aspersum Müller, 1774) were assigned multiple continental origins, as they exhibit intercontinental geographic distributions in their indigenous ranges. In scenarios where continental origin was obscure or unknown, the species was removed from this analysis (i.e., all species in the genus Allopeas Baker, 1935, Gulella Pfeiffer, 1856, Laevicaulis Férussac, 1822, Opeas Alber, 1850, and Subulina Beck, 1837). To assess climatic associations of each species in its native and CONUS environments, we categorized species by the Köppen-Geiger climate classification system (Rubel et al. 2017). Georeferenced records collected via GBIF and historical literature designating indigenous range were separately gathered and projected in ArcMap v.10.7 by ESRI. These records were spatially joined with a high resolution Köppen-Geiger climate zone projection (Rubel et al. 2017; available at koeppen-geiger.vu-wien.ac.at/present.htm) and the climate classification was extracted to each individual record. Similarly, we repeated this process for our curated database of comprehensive CONUS records to classify introduced climate association. Many species occupy multiple climate zones and are therefore included in all such occupied zones respective to each species. To visualize these data, the circlize package (Gu et al. 2014) was used in R v.3.6.1 (R Core Team). To enhance the interpretability of these visualizations, all classifications that yielded lower than ten observations were removed.

To assess spatial distribution through time of all non-native species, we projected records on a map of the contiguous U.S. at five time intervals starting from the first georeferenced record: 1862–1940, 1941–1960, 1961–1980, 1981–2000, and 2001–2019. The initial, large interval was used due to sparsity of records from the first georeferenced record until the mid-20^th century, followed by a standard two-decade delimitation. To identify areas with many non-native species records, the Point Density tool in ArcMap v.10.7 was used with a circular neighborhood of 75 km at each respective time interval. All time intervals were standardized to a single density scale.

Species richness and number of records in CONUS were quantified by political state boundaries by spatially joining record location data to a polygon layer of the contiguous U.S. Additionally, records were assessed in association to contemporary land cover type and human population density. We used the 2016 National Land Cover Database (NLCD; available at https://www.mrlc.gov/data/nlcd-2016-land-cover-conus) through the U.S. Geological Survey (Yang et al. 2018). Human population density data were obtained through the 2018 U.S. Census from the U.S. Census Bureau (TIGER/Line shapefiles; available at https://www.census.gov/cgi-bin/geo/shapefiles/index.php). The raster files were converted to point data and spatially joined to a polygon layer of 0.5 km buffers created around each record location. Land cover type was consolidated into seven categories: agricultural, barren, developed, forest, herbaceous, shrub/scrub, and wetlands.

From all sources, we assembled a dataset comprising 13,311 records for 25 families, 59 genera and 93 species. Of these records, 10,097 records included georeferences (with 134 records georeferenced by the authors), and 9,297 records included temporal information and georeferences. The full georeferenced dataset was used to generate the final taxonomic list, containing 22 families, 48 genera and 69 species (see Table 1). The majority of non-native species records are in California (5,735 of 10,097), with 26 of the total 69 species documented represented within the state (Fig. 1). Non-native species richness is more evenly distributed across various states, with all but two states (Nebraska and South Dakota) with at least one non-native species record.

Figure 1.

Bar plots of records (Top) and species richness (Bottom) by CONUS state including the District of Columbia. Nebraska and South Dakota are excluded with zero occurrences in these states.

The most prevalent and widespread species documented is Cornu aspersum, with nearly three times as many CONUS records (3,672) as the next most prevalent species Otala lactea. Cornu aspersum records are densely clustered in metropolitan areas along the west coast (incl. California, Oregon and Washington) with many records in south-central Texas, the southern Midwest, and along the eastern seaboard (Fig. 2).

Figure 2.

CONUS distribution of the five most prevalent non-native species in relation to county-based U.S. population density.

The second most prevalent species, O. lactea (1,297 records) exhibits a similarly broad introduced distribution to C. aspersum, most commonly associated with coastal areas in the west (California, Oregon) and in the east (Florida). Additional records are clustered within the northeast (Michigan, New York, Vermont). Records of the third and fourth most prevalent species, Rumina decollata (998 records) and Limax maximus (756), are primarily within metropolitan areas along the west coast (e.g., Los Angeles and San Francisco, CA, Seattle, WA) and in the central U.S. (e.g., Dallas, TX). These major urban hubs appear to be hot spots for introduction of these terrestrial gastropods. The most geographically widespread non-native species was L. maximus, being found from coast to coast in 37 of the 48 states.

Of the four contributing source categories to all records, a large majority (7,917 of 10,097 records) are from the citizen science database iNaturalist. Museum and natural history collections contribute 2,131 records, state and federal governmental agencies contribute 24 records, and 25 records come from scientific literature not associated with museum collections.

Europe is the continental origin for the majority of non-native CONUS species identified, with 25 genera and 45 species with a strictly European origin. An additional ten genera and eight species have a broad Mediterranean distribution that encompasses Western Europe and Northern Africa (see Table 1). Proportionally, species indigenous to the aforementioned continental regions collectively make up 92.2% of CONUS records, and the remainder of species with certain origins come from Asia (2.8%), the Caribbean (0.06%), and Central and South America (0.01%) (see Fig. 3) . Of the species documented, Cornu aspersum records are the most widespread and numerous. This species has a Mediterranean distribution (and was accordingly categorized with both an African and European origin in Fig. 3) but given that the majority of species are being translocated from Europe, we infer that this species may be disproportionately transported from the northern extent of its native range. Thus, the frequency of introduction from northern Africa is likely to be proportionally smaller.

Figure 3.

Left: Relative contribution of each continental indigenous origin for non-native CONUS terrestrial gastropod species records. Right: Climatic associations of each non-native species in the indigenous range and it’s CONUS records utilizing the Köppen-Geiger climate classification scheme. Illustrations are subdivided by continent or a grouping of continents in relative proximity. Color codes are defined for each classification, and the two-letter code preceding each climate code identifies the respective region (AF = Africa, AS = Asia, CA = Central America, CB = Caribbean, EU = Europe, SA = South America, US = United States).

Climate zone associations in indigenous and CONUS ranges of most species were similar. Of the seven species reported from tropical climate zones in the Caribbean, Central America, or South America, all CONUS records were also associated with tropical or humid subtropical climates (largely found in southern Florida). Likewise, > 97% of CONUS records for the two introduced Asian species come from the same zone as their indigenous environment. All but two species (Lissachatina fulica Bowdich, 1822, and L. immaculata Lamarck, 1822) with indigenous ranges including Africa are associated with Mediterranean-influenced climates, although most of these species’ ranges also include several additional climate zones in Western Europe. As such, both the African and European fauna have a higher diversity of climatic associations in both their indigenous and CONUS ranges. However, there is significant overlap between the broad climate classifications, with the primarily temperate, Mediterranean, and boreal climate zones being the dominant associations for indigenous and CONUS ranges.

Land cover type associated with records of the non-native species identified is primarily developed (47.2%). Records are largely clustered around areas of high human population density and urban sprawl. Within the three states with the highest number of records (California, Texas, Florida, respectively), areas with rapidly growing recorded density are major cities. For example, 2,819 records are from Los Angeles County alone, which comprises over one fifth of our entire georeferenced dataset. Records not associated with developed land were generally evenly spread across the other major land cover type categories (see Fig. 4).

Figure 4.

Proportions of 2016 NLCD land cover type in relation to 0.5 km buffer surrounding each record.

Few CONUS introductions were discovered from the first record in 1862 until 1940 (see Fig. 5). From then on, the number of records roughly doubled every two decades until 1980. From 1981–2000, the increase in records declined to an estimated 30.8%. Overall, increased record density was primarily associated with eastern coastal states, with gradual extension into the central U.S. However, very likely owing to the advent of digital tools to record species observations (e.g., iNaturalist), in the past two decades there was an estimated 797% increase in non-native species records. Recorded density in the last two decades has been heavily centralized in western coastal states in areas of urban sprawl, as well as in urban hubs in central states such as Texas.

Figure 5.

Point density map of non-native species records at five different time intervals. High-density values were associated with 100 or more records within a 75 km circular neighborhood around each individual record. Records were cumulative for each respective interval and tallied on the right side of each map. Time series data associated with new species and records shown in bottom right.

Our results indicate that hot spots of gastropod introductions occur in highly urbanized areas. This generally conforms to previous findings showing a significant correlation, at several spatial scales, between introduced species diversity and human population size. Examples of this correlation include invasive plants (Campos et al. 2016; Vinogradova et al. 2018) and invasive animals (Spear et al. 2013). One driver of this association is that increasing human population density leads to increased importation (dispersal) of non-native species into an area via intentional and unintentional introductions. More people in an area inevitably leads to more opportunities for introduction of ornamental plants, weeds, pets and many other well-documented pathways of non-native species importation (Pimentel et al. 2005; Simberloff 2013). Another driver of this correlation is that more people in an area produce increased anthropogenic disturbances, altering native habitat which eliminates many native species and creates habitat for non-native species (McKinney 2001). There is also a likely artifactual contribution to this correlation: increasing human population densities create a sampling bias by increasing the likelihood that more species (including introduced and invasive species) will be observed and recorded (Barbosa et al. 2013). This is especially true given the rapid rise of citizen science programs and social media platforms, especially iNaturalist.

The geospatial analyses also show that introduction hot spots tend to occur in highly populated areas concentrated along coastal regions at several latitudes (Fig. 5). Comparable areas of human population density located away from coastal areas tend to have much lower non-native species diversity. This pattern conforms to findings that ports of entry are gateways to many introduced species, especially non-native horticultural plants (Jehlik et al. 2019) and animals such as invasive insects (Langor et al. 2009) and invasive terrestrial gastropods (Bergey et al. 2014) that are hitchhikers on such plants. A key implication is that these organisms might gradually disperse into the interior of the continent, as seen in Fig. 5. This could be accelerated by punctuated dispersal events, e.g., following applicable aforementioned mechanisms, that might also occur.

Our results also suggest that native to introduced range climate analogs are positive factors in non-native terrestrial gastropod diversity and pervasiveness. Most non-native terrestrial gastropod species found in our study are located in climate zones similar to their native ranges, e.g., species primarily of Mediterranean origin recorded in Pacific coastal states in the introduced range. Previous studies have documented evidence of such climate matching in other groups, including invasive fishes (Howeth et al. 2016) and reptiles and amphibians (van Wilgen et al. 2009). A study of European non-native land snails also found evidence of climate matching but with several important exceptions that demonstrated the importance of including (where possible) species traits as an explanatory variable in understanding non-native snail distributions (Capinha et al. 2014).

The contiguous U.S. (CONUS) harbors a greater non-native terrestrial gastropod diversity than other New World nations (Naranjo-García and Castillo-Rodríguez 2017; Darrigran et al. 2020). This may be the product of greater interest in malacology within the U.S., as well as the popularity and accessibility of citizen science media. However, there are still notable sampling gaps within the country, and therefore this representation probably underestimates the full taxonomic scope of non-native terrestrial gastropods. Our results corroborate findings of Dawson et al. (2017), that the majority of species richness and abundance records were found in predominantly coastal areas. This is also supported by a recent study of non-native mollusks in South America (Darrigran et al. 2020).

Although most non-native terrestrial gastropod species exhibit climate matching to their indigenous ranges (discussed above), there is notable variation in the extent of occurrence and abundance of records between species. While analogous climate conditions might thus promote successful introductions of terrestrial gastropods in CONUS or other areas, there are clearly other factors driving the success of some non-native species relative to others. Generalist characteristics and broad thermal tolerances might contribute to survivability in a new habitat (McKinney and Lockwood 1999). Also, other life history traits and reproductive strategies in terrestrial gastropods might facilitate establishment from small populations (i.e., hermaphroditism, large clutch sizes; Robinson 1999). For example, there is a substantial literature attributing such traits to Cornu aspersum, the most widely recorded species in our dataset (Guiller et al. 2012; Gaitán-Espitia et al. 2013; Nicolai et al. 2013; Nespolo et al. 2014).

The pet and aquarium trade, increasing trade in ornamental and agricultural plants, as well as human food preferences have contributed to the importation and spread of invasive terrestrial gastropods within the contiguous U.S. Though our findings cannot directly quantify the relative importance of each of these dispersal vectors, there are apparent correlations between the geographical abundance of records for particular species and likely mechanisms. For example, C. aspersum and Otala lactea are among the most common land snail species used in human food consumption (escargot) owing to their fast reproductive rates and high nutritional content (Dragićević and Baltić 2005). These species are found in high densities in major U.S. cities such as San Francisco and Los Angeles, CA, Portland, OR, and Dallas, TX. Increased demand for exotic dishes in such communities, some with high socioeconomic areas, might provide greater opportunity for escape and persistence (i.e., via high propagules pressure) compared with less cosmopolitan areas. Other idiosyncratic drivers of success may be at play when considering the pervasiveness of these non-native species.

We did not consider impacts of any non-native species in this study, and therefore cannot directly infer potential economic or ecological harm associated with our results. The invasiveness and deleterious impacts of many of these species have been comprehensively reviewed in other literature (e.g., Robinson 1999; Cowie and Robinson 2003; Cowie et al. 2009), and we encourage those interested to seek additional information about these topics elsewhere. Using our data, however, we can provide further utility to previous projections of select non-native species and their negative impacts (if any), specifically those of Cowie et al. (2009). This previous research sought to quantify the potential ‘pest significance’ of a variety of non-native gastropod species based on life history traits, propagule associations, invasion history, general ecology, and other biological and historical factors. Species considered (both aquatic and terrestrial in this case) were then scored individually and proportionally to all others within their dataset using these factors, with those scoring highest being projected as most ecologically harmful.

Of the non-native terrestrial species considered of high potential risk included in Cowie et al. (2009), we surprisingly note that few have a substantial number of records in our dataset. For example, the genus Cernuella was scored individually and proportionally highest among terrestrial groups in Cowie et al. (2009), yet our data collection yields only 29 records of this genus in CONUS associated with two species (Table 1). The most prevalent species in our dataset, C. aspersum, was ranked in the lower extent of the top one third of the nearly 50 species considered in Cowie et al. (2009), along with another fairly prevalent species in our dataset (Theba pisana). All other species in this top one third of their dataset have fewer than 20 records. We do note, however, that the majority of our records do not have associated estimates of abundance (and thus this was not considered in our study), so few records of any one species should not infer a lack of future ecological harm. Instead, it can be used to inform management efforts in areas with species considered potential pests.

While the spatiotemporal trends exhibited in our dataset are consistent with other studies of non-native taxonomic groups within the U.S. (e.g., Mangiante et al. 2018), it is important to acknowledge limitations in this study and provide caution about its interpretability. Citizen science has become an increasingly popular tool both in terms of scientific analysis and to connect the general populous with the scientific community (Follett and Strezov 2015). So much so, in fact, that it here represents a impressively significant portion of our final dataset. For example, the Los Angeles County Museum of Natural History has initiated a new mollusk-specific citizen science program called Snails and Slugs Living in Metropolitan Environments (SLIME; https://nhm.org/community-science-nhm/slime). There have been a number of published products associated with this program (e.g., Ballard et al. 2017; Vendetti et al. 2018), and the vast majority of records in Los Angeles County (and by extension many in our dataset) are a product of participants of this program via iNaturalist. Thus, while these efforts have been successful and prolific, we advise caution when using all data provided for geospatial modeling without first accounting for common sources of geospatial error (e.g., spatial autocorrelation, pseudo-replication).