Truly invasive or simply non‐native? Insights from an artificial crested newt hybrid zone

Introductions of non‐native species can pose serious threats to native populations and ecosystems. However, the impact of introduced species depends on intrinsic characteristics, local habitat conditions, and the interaction with native species. Case‐specific management strategies may therefore be required. Using phenotypic characters and molecular markers for species identification, we provide insights into an artificial hybrid zone between two closely related newt species, the native Triturus cristatus and the introduced T. carnifex, near Tübingen, south‐west Germany. Our analyses revealed a central Italian origin of the non‐native T. carnifex and suggested their sustained presence in the study area for at least six years, probably much longer. In some ponds, extensive hybridization with native T. cristatus was detected. However, we found no evidence for a displacement of the native species by its non‐native congener. The gradient from pure T. carnifex to pure T. cristatus currently extends over 7 km. A future expansion of the hybrid zone and swamping of a neighboring T. cristatus meta‐population appears unlikely under the local configuration of breeding ponds. We propose to monitor the hybrid zone using genetic markers for evaluating the direction and speed of gene flow, complemented by capture‐recapture studies to reveal trends in species‐specific population sizes. To protect the native T. cristatus, we recommend practitioners to maintain their habitats, for example, by preventing illegal release of gold fish, by counteracting early drying of the breeding ponds, and by regularly cutting back trees and shrubs along the shoreline.


| INTRODUCTION
Non-native species play a major role in global environmental change. Several recent studies emphasize the detrimental effects that non-natives can have on native ecosystems (Gallardo et al., 2016;Jernelöv, 2017;Lavergne et al., 1999;Molnar et al., 2008), and call for a rapid eradication to avert ecological and economic damage (Pluess et al., 2012). Up to 25% of all species that have been introduced outside their native range become invasive (Jeschke & Strayer, 2005;Williamson, 1997), meaning that they spread rapidly and reach dominance in the new environment (Valéry et al., 2008). The percentage of introduced species causing some sort of negative impacts on native systems is smaller, because their effects on native species and ecosystems can also be neutral or even positive. For example, non-native species can provide prey or habitat to endangered native species or may increase the stability of ecosystems under climate change (Cattau et al., 2016;Dickie et al., 2014;Lampert et al., 2014;Schlaepfer et al., 2011;Sogge et al., 2008). For the management of introduced species, it may therefore be appropriate to distinguish between "non-native", "invasive", and "harmfully invasive" species. We consider a species non-native to an area where it did not historically occur and where it arrived by human-mediated ways of dispersal. Among non-native species, invasive species are those acquiring a competitive advantage in the new environment, enabling rapid spread and population dominance (Valéry et al., 2008). Harmfully invasive species are invasives with a strong and predominantly negative effect on the invaded ecosystems.
Hybridization appears to be particularly common along contact zones between members of the Triturus genus, for example, the Northern crested newt, Triturus cristatus (Laurenti 1768), and the Italian crested newt, Triturus carnifex (Laurenti, 1768) (Wielstra et al., 2014). The natural hybrid zone between the two species is relatively narrow (approximately 15-30 km) with a predominance of pure parental genotypes at the edges and hybrids in the center (Maletzky et al., 2008;Mikulíček et al., 2012). Apart from their natural hybrid zone, the two species co-occur along artificial contact zones caused by human translocations of T. carnifex into the native range of T. cristatus. Such introductions have been reported from several European countries (UK: Brede, 2000;Brede, 2015;Griffiths, 1996, Switzerland: Arntzen & Thorpe, 1999Dufresnes et al., 2016;Dufresnes et al., 2019, the Netherlands: Bogaerts, 2002Meilink et al., 2015;Wielstra et al., 2016, andGermany: Franzen et al., 2002;Maletzky et al., 2008). These introductions deserve special attention by conservationists, because both species face a strong decline throughout their native ranges, and introgressive hybridization between the two species may potentially lead to "genetic pollution" (Butler, 1994). The effect may, however, not be bi-directional, because it appears that alleles mainly introgress from the native into the non-native gene pool in the wake of an outwardly expanding hybrid zone (Currat et al., 2008).
Based on their external morphology, T. cristatus and T. carnifex are relatively hard to distinguish, especially when hybrids are present (Maletzky et al., 2008;Brede, 2015; Figure 1). Several external morphological traits have been proposed for species discrimination (Arntzen & Wallis, 1994;Brede, 2000;Wolterstorff, 1923). External characters, however, turned out to be unreliable when backcrossing of hybrids with one of the parental species occurs, and genetic tools are therefore required for studying hybrid zones (Brede, 2015;Dufresnes et al., 2016;Wielstra et al., 2016).
Here, we study an artificial hybrid zone between T. cristatus and T. carnifex near Tübingen (Germany). The studied waterbodies cover a gradient from a forested site (100% forest cover within a radius of 100 m) and two sites at the forest edges (79% and 44% forest cover in the surroundings) to several ponds in an agricultural matrix (no forest cover in the surroundings). In the north-west of the study area, a meta-population of T. cristatus with high conservation relevance is located. To distinguish between species, identify hybrids, and examine the geographic origin of the non-native newts, we employ external morphological criteria commonly used for species identification and molecular markers consisting of microsatellites and mitochondrial DNA fragments (Canestrelli et al., 2012;Dufresnes et al., 2016;Dufresnes et al., 2019;Wielstra et al., 2013;Wielstra et al., 2021). We incorporate information on pond age to develop a likely scenario for the dispersal of crested newts in the study region.

| Study species
Triturus cristatus is distributed north of the Alps, spanning Great Britain and most of central Europe from western France to the Ural Mountains in the east (e.g., Thiesmeier & Kupfer, 2000;Wielstra et al., 2014). Typically, the species inhabits medium-sized waterbodies with dense aquatic vegetation and strong insolation (Thiesmeier & Kupfer, 2000). Triturus cristatus has a relatively slender build, a dark, warty skin, and a yellow to orange belly with sharply edged black ventral dots (Arntzen, 2003;Arntzen & Thorpe, 1999). In contrast, T. carnifex has a relatively stout body, a smooth grayish skin with only few white lateral stippling, and round, blackish, diffusely edged dots on the venter (Arntzen & Thorpe, 1999;Brede, 2000, Figure 1). Female, juvenile, and subadult T. carnifex can often be identified by a yellow vertebral stripe (Franzen et al., 2002). The native range of T. carnifex covers most parts of Italy and the southern Alpine area (e.g., Andreone & Marconi, 2006;Arntzen, 2003). Habitat requirements of both crested newt species are similar, although T. carnifex might thrive in a broader range of habitats, including anthropogenically impacted waterbodies (Arntzen, 2003;Arntzen & Thorpe, 1999;Meilink et al., 2015).
Native populations of T. cristatus and T. carnifex are legally protected under annexes II and IV of the European Habitats Directive (92/43/EEC, Arntzen, Kuzmin, et al., 2009;Romano et al., 2009). In Germany and in the state of Baden-Württemberg, T. cristatus is considered vulnerable and endangered in the national and regional Red Lists, respectively (Geiger et al., 2020;Laufer, 1999). Along a small contact zone in northern Austria and the Czech Republic, the native ranges of the two species overlap and hybrids were found (Maletzky et al., 2008;Mikulíček et al., 2012). The occurrence of native T. carnifex in the very south-east of Germany was suspected based on morphological criteria (Schmidtler, 1976). A recent molecular study confirmed the presence of T. carnifex alleles in this region, but no genetically pure individuals of T. carnifex were found (Fahrbach et al., 2021).

| Study area and sampling sites
Our study area is located near Tübingen,Germany (48.52 N,9.05 E, 345 m-469 m a.s.l.; Figure 2), and is characterized by a forested hill and an adjacent river valley. The area is ecologically well-studied and among the most species-rich regions in Germany (Gottschalk, 2019). Even if waterbodies are rare, eleven different native amphibian species were recently found (Bamann, 2019). Triturus cristatus, which had been quite abundant in former times, has drastically decreased in the south-eastern part of the study area, where it likely disappeared during the mid-1900s (Löderbusch, 1987;Schmid, 1966). A single crested newt record is available from 1987 (Straub, 2013; site B, Figure 2). More recently, crested newts were resighted in several ponds, and habitat management actions were set in place (Straub, 2013).
F I G U R E 1 Phenotypic appearance of typical male crested newts. Triturus cristatus (left), Triturus carnifex (right), and T. cristatus Â carnifex hybrid (middle). Note that hybrid phenotypes are highly variable and, if at all, are difficult to distinguish from parental phenotypes. The shape of the male newts' crest is not a suitable indicator of species membership.
We first observed putative T. carnifex at site B in May 2015. The site is located on top of the hill, enclosed by forest, and consists of one medium-sized old quarry pond (≈ 300 m 2 ) and a few small temporary waterbodies nearby. In contrast, sites A and C are located at the forest edges approximately 1.1 and 0.8 km to the east and north-west from site B, and sites D-H are embedded in an agricultural matrix ( Figure 2). We started collecting genetic samples in 2016 at sites B, D, and H, for which the presence of crested newts was known. During 2017-2021, we checked a further 16 ponds for the presence of crested newts. We collected genetic samples at sites A, C, and E in 2017-2019. A few additional samples were collected from site B in 2017 and 2018, and site D was sampled again in 2019. In 2021, we surveyed sites F and G for the first time and morphotyped the crested newts from these sites without collecting genetic samples. The surveyed ponds include all potentially suitable breeding ponds within a radius of 5 km from site B, the nearest known population of pure T. cristatus (site H, Figure 2), and all waterbodies (except for garden ponds and private fish ponds) that could potentially serve as "stepping stones" from pure T. carnifex or admixed populations towards the nearest T. cristatus site.

| Field methodology
Surveys of crested newts were conducted with funnel traps (floating, 47 Â 23 Â 23 cm) and dip-netting. Based on phenotypic criteria, we distinguished T. cristatus, F I G U R E 2 Occupied and potential crested newt sites (Triturus cristatus/T. carnifex) and interpond distances in the study area. Sites that were considered largely unsuitable for crested newt breeding include those with fish and those strongly shaded, or heavily anthropogenically impacted. We present the number of individuals that were sampled for genetic analyses between 2016 and 2019 (n) and the genetic ancestry of the populations. A mean ancestry of 0 indicates a population of pure T. cristatus, 1 indicates pure T. carnifex, and values in between represent admixed populations. At site A, four pure T. cristatus were included in the genetic sampling. However, we observed phenotypic T. carnifex at the site a few weeks before our genetic sampling campaign. No genetic samples were available from sites F and G. Note the gradual transition from non-native T. carnifex at site B, over admixed populations at sites C-G, towards native T. cristatus at site H.
T. carnifex, and hybrids in the field. We refrained from species identification of larvae using phenotypic traits, because no clear external characters are available to discriminate between the species in early to mid-aged larval stages. Phenotypic species classification was later compared to genetic classification.
A total of 101 crested newts from six different sites were sampled for genetic analyses. Except for 18 individuals (15 larvae/juveniles and 3 adults from ponds A, B, and C), which were preserved as museum specimens, newts were sampled noninvasively using skin swabs (Prunier et al., 2012). We used conventional cotton buds to swab newts several times gently over the back, flanks, and tail. After swabbing, cotton buds were stored in SDS buffer or immediately dried with silica gel and brought to the laboratory.

| Laboratory procedure and data analysis
DNA was extracted from cotton swabs using standard protocols, including a proteinase-K treatment and filtration with QIAamp ® spin columns (Qiagen, Hilden, Germany). For cotton buds stored in SDS buffer, we modified the protocol by refraining from using the buffer ATL, adding the proteinase K directly into the SDS buffer, and consequently increasing the amount of the reagents (proteinase K, buffer AL, ethanol) threefold.
Following the procedure outlined in Dufresnes et al. (2016), we genotyped the sampled newts with nuclear (nDNA) and mitochondrial (mtDNA) genetic markers, allowing the detection of both current and historical hybridization events. We used a set of 11 microsatellite loci (nDNA) and one marker of the mitochondrial control region (CR, primer pair L-CR-Uro/H-tRNAPhe-Uro; Dufresnes et al., 2016). The microsatellite primers were originally designed for T. cristatus and/or T.carnifex and have proven useful for studying hybridization between the two species (Dufresnes et al., 2016;Dufresnes et al., 2019;Krupa et al., 2002). Mitochondrial genes have a strong tendency to introgress, in the wake of an outwardly expanding hybrid zone mainly from the native into the non-native species (Currat et al., 2008). This means that native mtDNA is usually retained in the gene pool of a population, even if the native nDNA may become entirely replaced ("cyto-nuclear discordance," Toews & Brelsford, 2012). To infer the geographic origin of the non-native T. carnifex, we additionally sequenced the mitochondrial ND4-marker (primer pair ND4carnF1/ ND4carnR2, Canestrelli et al., 2012; Table 1) in eleven newts with mtDNA of T. carnifex, six individuals from site B, four from site C, and one from site D.
Microsatellite polymerase chain reactions (PCRs) were carried out in three multi-and one singleplex reactions with a volume of 10 μl (Table 1). Fluorescent dyelabeled primers were prepared according to the manufacturer's instructions and combined for multiplex PCRs. Each reaction contained 5 μl of Qiagen 2 Â Master-Mix, 2 μl of primer mix, 1 μl of PCR-grade H 2 O (0.8 μl of H 2 O + 0.2 μl of BSA in the singleplex and the multiplex 2 reaction), and 2 μl of template DNA. The temperature profile for the PCRs included a denaturation for 5 min at 95 C, followed by 10 cycles of annealing with t a being the primer-specific annealing temperature: 30 s at 95 C, 90 s touchdown from (t a + 5 C) to (t a + 0.5 C), and 30 s at 72 C. The extension step was performed with 28 cycles (25 for multiplex 3) with 30 s at 95 C, 90 s at t a , and 30 s at 72 C, followed by 30 s at 60 C.
For the mtDNA-analysis, amplicons were amplified in two separate 25 μl PCRs with 2.5 μl of 10 Â Qiagen PCR buffer, 2.5 μl of 25 mM MgCl 2 , 3 μl of dNTPs, 1 μl of BSA, 1 μl of each primer (10 pmol/μl), 2 μl of template DNA, 0.15 μl of Taq polymerase, and 11.85 μl of PCRgrade H 2 O (CR) / 0.17 μl of Taq polymerase, and 11.83 μl of PCR-grade H 2 O (ND4). After an initial denaturation for 3 min at 94 C, PCRs included 35 cycles with 30 s at 95 C, 30 s at 55 C, and 30 s at 72 C, followed by a final step with 72 C for 5 min. Clean-up of PCR-products was performed using the Promega Wizard ® DNA clean-up system (Promega, Fitchburg, US) and amplicons were sequenced with the forward primers and Big Dye v. 3.1 reaction chemistry. All amplicons were run on an ABI 3730 48-capillary sequencer at the LMU Sequencing Service, Munich.
We aligned sequences of 312 bp (CR) and 555 bp (ND4) with publicly available sequences from NCBI Gen-Bank ® . Haplotypes were called based on full matches of the candidate sequences with the references. Microsatellite alleles were scored using GeneMarker ® V2.4.0, and manually checked. We used GenePop in R (R Core Team, 2007;Rousset, 2008) to check for Hardy-Weinberg equilibrium and linkage disequilibrium in the pure T. carnifex population (site B) and nine reference populations with >5 sampled T. cristatus. P-values were adjusted with the Benjamini-Hochberg correction. The presence of potential null alleles was assessed with Micro-Checker (van Oosterhout et al., 2004).
To identify pure parental individuals and hybrids from the genetic data, we used three different approaches, that is, Bayesian cluster assignment with STRUCTURE 2.3.4 (Pritchard et al., 2000) and NewHybrids Version 1.1 beta (Anderson & Thompson, 2002) as well as coefficients of ancestry and interclass heterozygosity (Fitzpatrick, 2012). In STRUCTURE, we tested 1 to 16 genetic clusters (K) with 100,000 iterations after a burn-in of 10,000.
The maximum number of tested clusters (K = 16) corresponds to the total number of sampled ponds from within and outside the hybrid zone. We assumed correlated allele frequencies and the admixture model. We carried out ten replicate runs per K and determined the most likely K with the deltaK-method in Structure Harvester (Earl & von Holdt, 2012;Evanno et al., 2005). For this particular K, we then repeated the STRUCTURE analysis in 25 independent replicates with 500,000 iterations each, after a burn-in of 50,000. We combined the replicates with Clumpp (Jakobsson & Rosenberg, 2007) and displayed ancestry coefficients graphically using the R-package strataG (Archer et al., 2017). Furthermore, individuals were assigned to six early-generation hybrid classes (parental species, first and second generation hybrids) using NewHybrids with 500,000 iterations after a burn-in of 50,000. Using the "z-option", we invoked 163 individuals from 10 nearby populations (< 65 km from the study area) as references for T. cristatus and the individuals from site B (all identified as pure T. carnifex by STRUCTURE) as references for T. carnifex. We manually calculated coefficients of ancestry and interclass heterozygosity from a set of eight species-diagnostic markers (all but A8).
Thereby, we considered an allele diagnostic for one species when it was frequently observed in that species but had a maximum of two occurrences in the other species. Species-diagnostic alleles for T. cristatus and T. carnifex were coded with 0 and 1, respectively. We averaged ancestry and heterozygosity coefficients across loci such that ancestry coefficients of 0 (T. cristatus) and 1 (T. carnifex) indicated pure parental genotypes and heterozygosity coefficients of 0 and 1, respectively, indicated a pure parental genotype and full interspecies heterozygosity. To infer the autochthonous status of T. cristatus in the study area, we ran STRUCTURE on a subset of our data, containing only pure T. cristatus from the study region and from the 10 reference populations (K = 1-13, 100,000 iterations, burn-in of 10,000, 10 replicates).

| RESULTS
We obtained high-quality DNA of sufficient quantity permitting microsatellite analyses and the sequencing of two distinct mitochondrial DNA markers in 98 and 84 out of 101 sampled crested newts from six different sites within T A B L E 1 Nuclear and mitochondrial genetic markers used in this study, references, annealing temperatures (t a in C), and reaction names. We report ranges of fragment lengths and haplotypes observed in the reference populations and present the number of speciesdiagnostic alleles (n) for the microsatellite markers.  Figure 2). The genetic analyses confirmed that both crested newt species, that is, the native T. cristatus and the introduced T. carnifex, co-occur in the study area and hybridize in several ponds. Interestingly, one newt, first captured in 2016 and recaptured in 2019, showed remarkable phenotypic plasticity, that is, a blackish dorsal coloration and a sharply edged belly pattern in 2016 but a more grayish skin and diffuse ventral dots in 2019 ( Figure 3). As a consequence, the exact same individual was phenotypically identified as T. cristatus in 2016 but as a hybrid in 2019. Genetic analyses suggested an F2-like nuclear genetic status for this individual. At sites F and G, two out of four and three out of eight captured individuals showed a hybrid phenotype, whereas the others were classified as T. cristatus.

| Nuclear genetic status
Data on the newts' nuclear genetic status were obtained from a set of 11 microsatellites. In a few instances, we detected more than two allele peaks in individuals from pond C. As the origin of the additional allele peaks was unclear, we removed the data from our analysis. The markers Tcri46 and D5 failed to amplify in our putative T. carnifex samples and were therefore also excluded. Finally, individuals with missing data in more than four out of the nine remaining markers were removed from the dataset. We observed significant deviations from Hardy-Weinberg equilibrium (HWE) in two markers (Tcri29: T. carnifex; A126: six out of nine tested T. cristatus populations) and potential null-alleles for Tcri35 (T. carnifex) and A126 (some T. cristatus populations). However, deviations from HWE and potential null-alleles were not consistent across populations; consequently we decided to keep the markers for the analysis. The marker D1 was monomorphic in the pure T. carnifex population and D127 was monomorphic, though species-specific, in both species. Tests for linkage disequilibrium did not indicate any significant associations between markers. We analyzed the microsatellite dataset with three different approaches and obtained highly concordant results. The STRUCTURE analysis with nine polymorphic microsatellite loci suggested a clear segregation into two clusters, representing introduced T. carnifex and native T. cristatus ( Figure S1). Ancestry coefficients from STRUCTURE were strongly correlated with those calculated from diagnostic loci (R = 0.99, p < .001). Only individuals with intermediate STRUCTURE ancestry coefficients were identified as hybrids, that is, F2 or backcross to T. carnifex, by NewHybrids (Figure 4).
The STRUCTURE analysis confirmed the morphological identification of crested newts as pure T. carnifex at site B and pure T. cristatus at site H. All three methods suggested extensive hybridization at sites C, D, and E with a high percentage of T. carnifex alleles at pond C and a predominance of the T. cristatus gene pool at site F I G U R E 3 Strong phenotypic plasticity in the belly pattern of a hybrid crested newt (Triturus cristatus Â T. carnifex). In 2016, the belly pattern showed sharply edged, black dots, whereas dots were diffusely edged and grayish in 2019.

D. While F2-like individuals and backcrosses to
T. carnifex were abundant, no F1-hybrids and no backcrosses to T. cristatus were observed at the studied ponds. "Pure T. cristatus" was the most likely early-generation hybrid category for all sampled newts from site H. Discriminating between six discrete hybrid classes is, however, likely to be an oversimplification because ancestryheterozygosity plots indicated a large variability in the hybrid genotypes at ponds C, D, (and E), suggesting hybridization over multiple generations (Figure 5). At pond D, where newts were sampled in 2016 (n = 8) and 2019 (n = 11), no significant shift of newt ancestry was detected within the three years (mean ancestry coefficient 2016: 0.269, mean ancestry coefficient 2019: 0.282, F 1,18 = 0.013, p = 0.912). At least some T. cristatus from sites A and D clustered with individuals from site H and other populations from central Baden-Württemberg ( Figure S2, Figure S3).

| Mitochondrial genetic status
The studied CR marker allowed a clear discrimination between the species. Within our study area, we found one CR haplotype each for T. carnifex (CAR3a) and T. cristatus (CRI2/2a/3). The two lineages differed clearly (8 bp). All sampled newts at sites A and H possessed the mtDNA of T. cristatus, whereas newts from site B exclusively carried mtDNA of T. carnifex. Mitochondrial DNA from both species was found at sites C and D, with a strong predominance of the T. carnifex mtDNA at pond C and the T. cristatus mitotype at pond D (Figure 4).
In a subset of eleven T. carnifex/hybrids from the study area, we consistently observed the ND4-haplotype CIII10/11. This haplotype has previously been reported from central Italy (Mulino di Pianoro, Emilia-Romagna; Minucciano, Tuscany; Canestrelli et al., 2012) and the F I G U R E 4 Phenotypic, mitochondrial, and nuclear genetic classification of 98 genetically sampled crested newts (Triturus cristatus/T. carnifex) from six different sites. Each column represents one individual. STRUCTURE and NewHybrids confirm the presence of pure T. carnifex at site B, pure T. cristatus at sites A, D, H, and hybrids at C, D, E. Phenotypically clear T. carnifex have been observed at site A prior to our sampling campaign. Substantial cyto-nuclear discordance was detected in only one individual from site D, bearing mtDNA of T. cristatus but predominantly nDNA of T. carnifex. Pure T. carnifex were phenotypically well separated from T. cristatus. However, in some instances, pure T. cristatus were erroneously considered hybrids from their phenotypes and hybrids were erroneously assigned to the parental phenotypes. Note the gradient from pure T. carnifex at site B over T. carnifex-and T. cristatus-dominated mixed populations, respectively, at sites C and D, towards pure native T. cristatus at site H. We refrained from a phenotypic classification for larvae. Limited amounts of template DNA did not allow mtDNA sequencing in a few individuals. non-native T. carnifex population near Basel (Dufresnes et al., 2019).

| DISCUSSION
Our study confirmed a new record of introduced Italian crested newts (T. carnifex) near Tübingen (Germany) with a central Italian origin. Their low allelic richness indicates that the population was founded by a limited number of individuals, theoretically by as few as a single breeding pair. Triturus carnifex and hybrids with the native Northern crested newt (T. cristatus) occur in the study region in several ponds.

| Phenotypic and molecular species identification
Nuclear and mitochondrial genetic markers confirmed the phenotypic species identification at sites with pure populations of either T. cristatus or T. carnifex. Conversely, at ponds with hybrids (sites C, D, E), phenotypic classification was only partially in line with the newts' genetic status. The genetic analyses confirmed the phenotypic classification as "hybrid" in 50% of individuals, whereas in the other 50% genetically pure T. cristatus were erroneously considered hybrids. In admixed populations, pure T. cristatus were correctly identified by their phenotype in only 40% of the cases. The newts that were erroneously considered T. cristatus turned out to be hybrids ( Figure 4). Even though we never misclassified genetically pure T. cristatus as T. carnifex and vice versa, the high degree of misclassifications observed in our study indicates that phenotypic traits are not useful for a reliable species identification in hybrid zones, confirming previous findings (Brede, 2015). The large differences in the number of morphoÀ/genotyped individuals between sites can be explained by differences in the local population sizes, differences in pond characteristics (e.g., water depth, vegetation cover) affecting capture probability, and different sampling effort.

| Spread and dominance
During the past century, several amphibian and reptile species were introduced into the study region (Bamann, 2019; Schmid, 1966). The site of introduction of T. carnifex and the pattern of spread cannot be inferred with certainty from our data. However, multiple pieces of evidence are in favor of the scenario that T. carnifex was originally released at site B or C and, while living undetected in the area for many years, its alleles have dispersed over 3 to 4.5 km.
Our data show a clear gradient from pure non-native T. carnifex to pure native T. cristatus within 7 km distance between sites B and H. In between (sites C, D, and E), crested newts showed variable degrees of admixture, specifically a gradually decreasing frequency of T. carnifex alleles with increasing geographic distance from site B. We found no alleles of T. cristatus at site B and therefore consider as factual that T. cristatus was absent immediately before the arrival of T. carnifex. At site C, we found mtDNA of T. cristatus in 1 out of 16 genotyped individuals and we consider it likely that T. carnifex also was the first crested newt at this site, while alleles of T. cristatus may have reached the population with immigrating T. cristatus and/or hybrids from site D or E. If a vital population of T. cristatus had been existing prior to the arrival of T. carnifex at site C, we would expect to still find the T. cristatus mitotype at high frequencies (Currat et al., 2008). Alternatively, strongly asymmetric mate choice, that is, preferably ♀ T. carnifex Â ♂ T. cristatus mating, or asymmetric incompatibilities could explain the observed pattern ). However, we observed many hybrids with the T. cristatus mitotype at site D, what makes this explanation rather unlikely.
Apart from the sites where T. carnifex has likely been the dominant crested newt species since its arrival, the species did not reach dominance at the studied waterbodies. Ancestry-heterozygosity plots clearly revealed extensive hybridization between the two species at sites C, D, and E for more than two generations. This means that alleles of both species occurred at the sites for at least six years prior to our study (see Cvetkovi c et al., 1996 for the generation time of T. carnifex). Our data do not support any factual statement going beyond six years, but it appears likely that T. carnifex has thrived in the study area for much longer. Schmid (1966) and Löderbusch (1987) systematically surveyed amphibian breeding ponds in the south-eastern part of our study area (including sites A, B, C, and E) but did not observe crested newts. Crested newts can be overlooked when present in low densities and when standard monitoring methods such as dip-netting are used (e.g., Kupfer, 2001). Nevertheless, the lack of records despite two independent amphibian surveys points to the absence of crested newts from this part of the study area during the 1960s to early 1980s. The first more recent record dates back to 1987, when Straub (2013) observed a crested newt (unfortunately without determining the species) at site B. It seems well possible that this individual was one of the first T. carnifex being introduced, because sites E-G have been only constructed after 1980constructed after (E: 1983constructed after , Löderbusch, 1987F: 1987, Kienzler pers. communication;G: 2018, Stoltze pers. communication) and site D is at least not visible on aerial images dating back to 1968. It is therefore hard to imagine from where T. cristatus could have immigrated into site B. Four probably suitable breeding ponds for the species during the mid-1900s were in the surroundings of site H, 7 km apart from site B. Three of the waterbodies still exist and are nowadays part of a T. cristatus metapopulation (Figure 2).
Considering the lifetime dispersal capacity of T. carnifex of approximately two kilometers (Mori et al., 2017) and its likely presence in the study area for a couple of years, it is not surprising that its gene pool has spread. Our data suggest that both crested newt species have (re-)colonized the south-eastern part of our study region within the past 30-40 years, meaning that both species and/or their hybrids may occasionally disperse 2.2 km (inter-pond distance C-E). Relatively strong dispersal of introduced T. carnifex was reported from the Veluwe (Netherlands, Meilink et al., 2015), from the Geneva Basin (Switzerland) after multiple independent releases (Arntzen & Thorpe, 1999;Dufresnes et al., 2016), and from São Miguel (Azores) where congeners are absent and T. carnifex was released more than 100 years ago (Machado et al., 1998;Svanberg, 1975). In contrast, the population at Beam Brook (UK, Brede, 2015) has not spread considerably in the 80 years after its introduction and the hybrid zone near Basel (Switzerland) is currently restricted to a few ponds within two kilometers from the putative introduction site (Dufresnes et al., 2019).

| Impact on native species
In contrast to Meilink et al. (2015) and Dufresnes et al. (2016), who reported strong introgression of native mtDNA into individuals with a predominance of nonnative nDNA, we did not observe substantial cyto-nuclear discordance. Interestingly, all genetically sampled newts from site A (including three subadult newts) were T. cristatus, although we observed phenotypic T. carnifex a few weeks before our genetic sampling campaign (Bamann, 2019). We hypothesize that T. carnifex has either arrived very recently at the site or that the two species have clear assortative mating preferences (see Michalak & Rafínski, 1999 for Lissotriton), because a considerable introgression could be expected even with a very low frequency of interspecies mating (Currat et al., 2008). At pond H, all individuals were identified as T. cristatus in NewHybrids, nevertheless the hybrid index suggested some introgression of T. carnifex alleles for two individuals. This may result from either genotyping errors, imperfectly species-diagnostic alleles, or true gene flow.
Apart from genetic replacement, native T. cristatus may potentially suffer from a loss in fitness after hybridization with the non-native congener (e.g., Muhlfeld et al., 2009). Although hybridization has apparently no effect on the survival of larvae (Wyssmüller, 2007), it remains unclear how different degrees of introgression may affect the fecundity of crested newts and the hatching rate of embryos (compare to Vucic et al., 2020). Yet, disrupted meiosis and dysfunctional gametes were reported for hybrids (Callan & Spurway, 1951), and relatively many malformations were observed in a natural hybrid zone of the two species (Mač at et al., 2015). We observed foot malformations (missing and supplementary toes) in two phenotypic T. cristatus from site G. However, the degree of malformations at our study sites is not significantly increased compared to >10 other T. cristatus populations that we studied systematically since 2016 (Hinneberg, unpublished data) or to values from the literature .
When the phenotype of the non-native species and the hybrids is significantly different from the native phenotype, hybridization can have severe impacts on "third-party community members" (Ryan et al., 2009), that is, species that are not genetically affected by hybridization. Crested newts may occasionally prey on other amphibians, in particular their eggs and larvae. However, amphibians are not the main prey items for either T. cristatus or T. carnifex (Careddu et al., 2020;Dolmen & Koksvik, 1983;Fasola & Canova, 1992;Iannella et al., 2020;Roşca et al., 2013), meaning that it may make little difference to other amphibians whether the native or the non-native species, or hybrids co-occur in the pond. The limited impact of T. carnifex and hybrids on native amphibians may also be evidenced by the rich amphibian fauna at our study sites, with up to seven species per pond, including threatened species such as Bombina variegata and Pelophylax lessonae.

| Invasiveness and future invasive potential
Recently, T. carnifex was considered "one of the most successful invasive amphibians in Europe" (Dufresnes et al., 2019). However, although introductions of T. carnifex may be relatively frequent (Arntzen & Thorpe, 1999;Bogaerts, 2002;Brede, 2000;Brede, 2015;Dufresnes et al., 2016, Dufresnes et al., 2019Franzen et al., 2002;Machado et al., 1998;Maletzky et al., 2008;Malkmus, 1995;Meilink et al., 2015;Svanberg, 1975; this study), the appraisal "invasive" may need to be reconsidered (see also Brede, 2015). Valéry et al. (2008) characterize an invasive species by "a competitive advantage following the disappearance of natural obstacles to its proliferation", which enables the species to spread rapidly and become dominant. Our genetic data suggest that dispersal distances of introduced T. carnifex are similar to distances reported from the species' native range (Mori et al., 2017) and similar to dispersal distances of local T. cristatus. T. carnifex did not become dominant at any site that has previously been occupied by T. cristatus. Hence, our data do not provide evidence for a replacement of the native species. At pond D, where T. carnifex has likely immigrated into an existing population of T. cristatus, we observed extensive hybridization but still a dominance of T. cristatus alleles. Our current data do therefore not provide any support for T. carnifex being an "invasive" or even a "harmfully invasive" species and we propose to term T. carnifex in our study region "non-native." Nevertheless, we acknowledge that our study can only provide very limited information about the speed and the direction of gene flow and a follow-up study would be necessary to complement our snap-shot, also because predicting future trajectories of introduced populations is difficult and a certain time lag without signals of an invasion can be followed by population explosions (Crooks & Soulé, 1999). Visser et al. (2016) claimed that the lack of waterbodies serving as "stepping stones" can lead to the stabilization of a moving hybrid zone. Furthermore, they suggested that species-specific habitat preferences may shape hybrid zone dynamics. Both apply to the situation in Tübingen. There are currently no suitable breeding ponds for crested newts in the center of our study area and a small river may impede the dispersal of T. carnifex towards the northwest ( Figure 2). Consequently, the chance of T. carnifex alleles reaching site H and adjacent waterbodies without human assistance is small. In turn, the preference of T. cristatus for open habitats could prevent its immigration into site B. Therefore, we predict the hybrid zone to stabilize approximately at its current extent.

| Management implications
In case that alleles of T. carnifex may nevertheless reach the T. cristatus meta-population in the north-west by single immigrants, high-density blocking would likely hamper their establishment as long as the non-native alleles do not provide a selective advantage (De Meester et al., 2016;Waters et al., 2013). A capture-recapture study at site H suggested a population size of several hundred T. cristatus. In contrast, the population of pure T. carnifex at B was estimated at approximately 70 individuals in 2016 (Hinneberg, unpublished data).
Under the precautionary principle, it may be desirable to remove any non-native plant and animal population, irrespective of its current "invasiveness" to prevent future invasions (Pluess et al., 2012). Removing aliens is, however, often difficult, involves a certain risk of failure, and can have unintended side-effects (Pluess et al., 2012;Sogge et al., 2008). Considering the time to reach maturity for individual newts and the potential presence of "skipping breeders" (Cvetkovi c et al., 1996;Hagström, 1979), attempting removal of non-native T. carnifex alleles would require extensive trapping at all sites occupied by T. carnifex or hybrids and during a minimum of four consecutive years. Moreover, an eradication attempt could be harmful to both non-native and native crested newts, because hybrids cannot be reliably differentiated from pure native T. cristatus without determining the genotype of each individual newt (Brede, 2000, this study). Instead, we recommend a targeted monitoring of the hybrid zone, including genetic tools, and we encourage the use of skin swabbing for DNA sample collection. Limited DNA quantities were a major challenge in our study. To ensure sufficient DNA quantity and quality in future studies, we propose swabbing multiple times with a gentle pressure over the newts' back, tail sides, and venter, to dry the cotton tip immediately after swabbing with silica gel and to extract the DNA within four weeks after swabbing. We suggest an elution volume of 30-50 μl. Furthermore, we recommend the assessment of (species-specific) population sizes for all waterbodies in the study region with capturerecapture. The monitoring should be repeated in regular time intervals, for example, after 5-8 years (approximately two newt generations), thus allowing speed and direction of gene flow to be determined and trends in species-specific population sizes to be detected. To protect the native T. cristatus, we suggest specific measures of habitat maintenance at sites which are currently unaffected by hybridization but also at sites with a low frequency of non-native alleles. For example, managers should prevent the illegal release of gold fish, counteract the drying out of breeding ponds before newt larvae reach metamorphosis, and cut back trees, shrubs, and reeds that shade the waterbodies. A vital population of T. cristatus is key to the species' conservation and very likely to be the best protection against a future, potentially more harmful invasion of T. carnifex.

AUTHORS' CONTRIBUTIONS
Heiko Hinneberg conducted the fieldwork, the laboratory work, and the data analysis. Heiko Hinneberg contributed to the acquisition of funding and led with writing the original manuscript. Thomas Bamann contributed to fieldwork and provided funding. Julia C. Geue substantially contributed to the development of the laboratory protocol and supported laboratory work. Katharina Foerster contributed to the design of the study and provided funding. Henri A Thomassen supervised the laboratory work and supported data analysis. Alexander Kupfer contributed to the design of the study, supported fieldwork, and provided funding. Henri A Thomassen and Alexander Kupfer supported writing of the original manuscript. All authors critically revised the manuscript. Henri A Thomassen and Alexander Kupfer should be considered joint senior author.

ACKNOWLEDGMENTS
We thank D. Lüdtke and M. Raichle for their assistance in the laboratory as well as N. and M. Kaczmarek for their help with the fieldwork. Data collection was partially funded by the "Stiftung Artenschutz", the "Gemeinschaft Deutscher Zooförderer e.V.", and the "Stiftung Landesbank Baden-Württemberg". We would like to thank Ben Wielstra, the associate editor Carly Cook, and two anonymous reviewers who provided valuable comments on an earlier version of the manuscript. Open Access funding enabled and organized by Projekt DEAL.

CONFLICT OF INTEREST
We declare no potential conflict of interest.

DATA AVAILABILITY STATEMENT
The full dataset, containing phenotypical and genetic data, is accessible as supporting information.