Predicting successful invaders is a central theme for invasion ecology, yet only three factors yield consistent associations with establishment success across regions and taxa: climate match, prior invasion success, and propagule pressure (Kolar and Lodge 2001; Hayes and Barry 2008; Gallien and Carboni 2017). Of the three consistent predictors, climate match is the most fundamental because some degree of suitability is necessary for establishment (Hayes and Barry 2008; Olyarnik et al. 2009). Climate therefore provides a strong primary invasion filter (Chapman et al. 2014), reducing the pool of potential invaders to those species capable of surviving, reproducing, and spreading within the regional constraints of seasonal variation in temperature and precipitation.

Climate match is frequently associated with establishment success in freshwater fishes. An analysis of 280 species in 10 countries identified a simple model using climate match and invasion history to correctly categorize 78% of successfully established species (Bomford et al. 2010). The mean climate match for successfully established non-natives was greater than the mean climate match for failed introductions for each of the countries in their study (Bomford et al. 2010). For the heavily invaded Laurentian Great Lakes, climate match alone was predictive of establishment success with 75–81% accuracy (Howeth et al. 2016). This past success in incorporating climate match increases the confidence that risk managers can place in assessments (Hayes and Barry 2008).

Most regional studies assessing fish establishment success focus on temperate or cold climates of North America or Europe (Garcia-Berthou 2007), and global analyses have explained little variance in establishment success (e.g., 12%, Ruesink 2005) or have included few warmer climate locations (e.g., Bomford et al. 2010). More thorough investigation of warm climate regions is warranted to determine the consistency of climate match as a predictor of establishment. A broadening of geographic scale is needed in response to the growing numbers of freshwater fish introduced into tropical and warm temperate climate zones in Mexico (Espinosa-Pérez and Ramírez 2015), Central and South America (Britton and Orsi 2012; Esselman et al. 2013), Florida (USA) (Shafland et al. 2008; Robins et al. 2018), Africa (Ellender and Weyl 2014), south and southeast Asia (Arthur et al. 2010; Herder et al. 2012), and Australia (Lintermans 2004). Such studies can lead to better understanding of the complex process of invasion across geospatial scales and taxonomic diversity.

Florida (USA) is an important region for testing hypotheses related to non-native freshwater fish establishment, with at least 122 species reported, of which 48 species have achieved persistent, reproducing populations (Robins et al. 2018). Peninsular Florida has multiple invasion pathways, hotspots of dense human population, and abundant and diverse aquatic habitats, all factors thought to increase its vulnerability to non-native fish introduction and establishment (Hardin 2007). Importantly, the climate of peninsular Florida is considerably different than the rest of the continental United States. It has some of the warmest winters in the warm temperate climate zone (Cfa) and the only tropical zones (Af, Am, and Aw) in the region (Beck et al. 2018). The goal of our study was to test the accuracy of climate match in a warm climate region for distinguishing successful versus failed introductions using the introduced freshwater fish fauna of peninsular Florida. Our specific objectives were to (1) test for mean differences in climate match between non-native fish species that have successfully established (hereafter, ‘successful’) and those that have failed to establish (hereafter, ‘failed’) using existing protocols (Bomford et al. 2010; USFWS (United States Fish and Wildlife Service 2020. Standard operating procedures: How to prepare an “Ecological Risk Screening Summary”. https://www.fws.gov/fisheries/ANS/pdf_files/ERSS-SOP-February2020-FINAL.pdf; hereafter, ‘USFWS SOP’) and (2) to determine if the climate match categories of the USFWS SOP could distinguish between successful and failed species. The results of this study can be used to evaluate the predictive ability of these climate match protocols and hence their utility for risk screening and assessment of potentially invasive species.

We developed lists of established and failed non-native freshwater fish species for peninsular Florida using a wide range of sources, including the U.S. Geological Survey’s Nonindigenous Aquatic Species database (USGS 2023. http://nas.er.usgs.gov/), published literature (e.g., Shafland et al. 2008; Schofield and Loftus 2015; Robins et al. 2018), field collections of the authors, and consultation with colleagues. Species were excluded from both lists if they were native transplants except for the Rio Grande Cichlid Herichthys cyanoguttatus Baird & Girard, 1854, a species native to southern Texas but universally included in such lists (e.g., Schofield and Loftus 2015), or if the introductions were outside of peninsular Florida. Species were categorized as successful if they had one or more established or reproducing populations in the region. Failed species included formerly reproducing species or those reported without evidence of reproduction; however, species were excluded if all known populations were eradicated by humans or if they represented introductions of a single individual (Lawson and Hill 2021, 2022). The resulting list included 37 successful and 36 failed species.

We estimated climate match in two ways. First, following a test of climate matching for the Laurentian Great Lakes region (Howeth et al. 2016), we used CLIMATCH (ABARES (2020) Climatch v2.0 user manual. Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. https://climatch.cp1.agriculture.gov.au/) and a U.S. Fish and Wildlife Service (USFWS) protocol for Ecological Risk Screening Summaries (ERSS; USFWS SOP) that incorporates all native and non-native established populations as source data, including locations within the risk assessment area. This procedure results in a post hoc determination of climate match for successful species. Secondly, we ran the same analysis using CLIMATCH but omitted all Florida locations (Bomford et al. 2010) to determine an a priori climate match, a more useful scenario for risk assessment (i.e., determining if a species not already introduced might establish). CLIMATCH, a simple freely available web-based application, has become the climate-matching program of choice for many risk screening activities in the United States and worldwide (ABARES (2020) Climatch v2.0 user manual. Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. https://climatch.cp1.agriculture.gov.au/; Froese 2012). The USFWSERSS previously utilized the program but has since developed a similar application for internal use (USFWS (United States Fish and Wildlife Service) (2019) Standard operating procedures for the Risk Assessment Mapping Program (RAMP). U.S. Fish and Wildlife Service. https://www.fws.gov/fisheries/ANS/pdf_files/RAMP-SOP.pdf). Both programs use temperature and precipitation data from land-based weather stations to determine similarity between a designated source region and a target region. As recommended in the protocols, we used the default set of 8 temperature and 8 precipitation variables (Suppl. material: table S1).

CLIMATCH analyses were completed for all species using source populations with (post hoc) and without (a priori) Florida locations for successfully established species. Location data were acquired through the Global Biodiversity Information Facility (GBIF 2023 https://www.gbif.org/), the U.S. Geological Survey Nonindigenous Aquatic Species database (USGS 2023. http://nas.er.usgs.gov/), taxonomic guides, and primary literature and assessed for accuracy before use. The target region was peninsular Florida, the part of Florida south and east of the Suwannee River system. The output of CLIMATCH includes similarity values for each weather station that range from 0 to 10, where 0 indicates no similarity and 10 indicates complete similarity (Bomford et al. 2010; USFWS SOP). The proportion of stations with a similarity of 6 or greater (Climate 6 Score = (Sum of counts for Climate Scores 6–10)/(Sum of all Climate Scores)) is the critical value, with values ≤ 0.005 indicating low climate match, those > 0.005 but < 0.103 indicating medium climate match, and values ≥ 0.103 indicating high climate match (Table 1; USFWS SOP). We tested for mean differences in climate 6 scores (1) between successful species (including Florida source locations) and failed species, and (2) between successful species (excluding Florida source locations) and failed species, using Wilcoxon Rank Sum tests in JMP Pro (V. 17.0, SAS Institute Inc., Cary, NC, USA). See Suppl. material: table S2 for species categories and Climate 6 scores and Suppl. material: table S3 for notes on species categorization.

With Florida locations included (post hoc), mean climate match (± SE) of successful species (0.991 ± 0.004) was greater (χ² = 18.88, df = 1, P < 0.0001) than the mean climate match for failed species (0.574 ± 0.072; Fig. 1). Climate 6 scores for successful species ranged from 0.87 to 1 and failed species from 0 to 1. Of all introduced species with Climate 6 scores ≥ 0.87, 72.6% were successful. In contrast, without Florida source locations (a priori), the mean climate match (± SE) for successful species (0.647 ± 0.062) was not significantly different (χ² = 0.0436, df = 1, P = 0.834) from the mean climate match for failed species (0.574 ± 0.072). In this latter case, Climate 6 scores for successful species ranged from 0 to 1. On average, the inclusion of data from the target region inflated climate match scores 1.5 times (0.991 vs. 0.647).

All Climate 6 scores for successful species in the post hoc analysis were equal to or greater than those in the a priori analysis. Climate 6 scores for 11 successful species did not differ between post hoc and a priori analyses whereas Climate 6 scores of the remaining 26 successful species differed by values (Δ) ranging from 0.012 to 0.991 (Fig. 2). Mean Δ was 0.343 (SD = 0.365) for all 37 successful species. The most extreme example of the variation in post hoc versus a priori scores is the goldline snakehead Channa aurolineata (Day, 1870), a species native to Southeast Asia (Fig. 3). Using occurrence data from southeast Florida, the post hoc Climate 6 score was 1.0. Conversely, leaving out Florida source data (a priori) resulted in a much lower Climate 6 score of 0.009. Therefore, the a priori analysis indicated relatively little similarity between climates in Southeast Asia and peninsular Florida whereas the post hoc analysis suggested considerably more potential range of climate match in the risk assessment area.

Post hoc Climate 6 scores for successful species resulted in all 37 being classified as having a high climate match (>0.103) and a priori scores resulted in 32 species with a high climate match, 3 with medium match, and 2 with low match (Table 1). For failed species, 29 species had a high climate match, 4 had a medium match, and 3 had a low match (Table 1).

We found little evidence for the association between a priori climate match and establishment success of non-native freshwater fishes in peninsular Florida, a warm climate zone. This finding is in contrast to the robust consensus in the literature that climate match is a consistent predictor of invasion success for a wide range of taxa (Hayes and Barry 2008), including freshwater fishes (Bomford et al. 2010; Howeth et al. 2016). Following a similar protocol using data from the target region led to differing results of climate match—strongly predictive post hoc versus non-predictive a priori of establishment success using the same data set. However, the post hoc analysis is of limited utility for prediction because it is tautological—the scores are high because the species is successful, and the species is successful because the scores are high. Thus, the climate match procedure may strongly influence the outcome of analyses. Our results have considerable implications for the use of climate match as a predictive tool for freshwater fish invasions, risk assessment in general, and management of potentially invasive species.

This work was funded by the University of Florida College of Agricultural and Life Sciences.

JEH, QMT, KML research conceptualization; JEH, QMT, KML sample design and methodology; JEH, QMT, KML investigation and data collection; JEH, QMT, KML data analysis and interpretation; JEH funding provision; JEH, QMT, KML writing - original draft; JEH, QMT, KML writing - review & editing.