Discussion
Several invasive species are known to have populations with low genetic diversity (Bai et al., 2012; Rollins et al., 2009; Santos et al., 2013). Besides cases where individuals from multiple native sources are put together (Kolbe et al., 2004), low genetic diversity and increased rates of inbreeding are expected due to sequential bottleneck effect (Estoup et al., 2011). Countries that have been colonized by bullfrogs not always received them directly from native sources, resulting in populations with low genetic diversity, which facilitates tracking their introduction pathway by genetic similarity. The association between genetic data and information from specific literature along with historical events can be helpful on understanding introduction pathways.
In our analysis, microsatellite and mitochondrial markers have shown partially consistent results. There is a consensus between markers that most areas of South America belong to the same genetic population. Two different haplotypes were found in this continent (Fig. 2). One of them (H01) is represented by 75% of the samples from Brazil and Ecuador. The other 25% is represented by H02, which is the same haplotype found among samples from Minas Gerais and Cuba on the cytb analysis. This result is consistent with the DAPC, in which Cuba has only a marginal link with the South American cluster. It also suggests that Ecuador received bullfrogs from Brazil after the secondary introduction from Michigan that might have happened before 1985.
Although no information about the native origin of the founders of Cuban population is available [besides brief and weak conjectures based on morphotype variation by Hoffman & Noble (1927)], it is expected that multiple events of introduction would increase genetic variation. Only one haplotype was found in Cuba, suggesting a single source of origin or a great loss of variability due to genetic drift since the last introduction event in 1946. In any case, it is reasonable to consider that samples from Cuba and Minas Gerais share their native source, as shown in the θST differentiation analysis. Both regions likely received animals from Michigan, since there are no records of a secondary introduction between Cuba and Brazil.
Given the records of close ties between Cuban and Chinese bullfrog lineages in the literature, one might expect them to exhibit same haplotypes, which would differ from those in Japan, since the only documented introduction from Japan to China failed. However, two haplotypes can be found in China. One of them is the same from Cuba, while the other is found in all sampled Asian countries.
Literature indicates Japan as the source population to all countries of Asia, besides China, and genetic data partially corroborates this information. The two haplotypes found in Japan are very similar to one another, with only one polymorphic locus, being recovered as sister groups in our phylogenetic analysis. This could either mean that they came from close areas of the native range or a single haplotype suffered a mutation after introduced to Japan. One could be understandably biased in favor of the second hypothesis, due to the lack of one of these haplotypes in the rest of Asia, meaning the mutation happened after Japan exported specimens to other countries. Unfortunately this is a tricky question with many possible answers, such as: 1. Japan received this haplotype from native source after exported animals to Taiwan; 2. Japan received this haplotype before exporting animals, but it was not sampled among the exported specimens; 3. Other countries received this haplotype from Japan, but it was lost due to genetic drift; 4. We did not sample this haplotype in other countries by casualty. Sampling in DPR Korea Ri (2018) may be a step towards answering these questions as these specimens have been introduced from Japan possibly before 1950, and thus before a potential mutation was exported to other countries.
In any case, the haplotype H03 is widespread in Southeast Asia, including China. Several different situations are likely to have happened to cause this genetic homogeneity in Asia. The most parsimonious hypothesis is that other translocations happened from Japan to China. There are few reports that it may have happened at different periods. Although no strong evidence is available, it is likely to have happened during the Japanese invasion of China at the second Sino-Japanese war in between 1937 and 1945, and another in 1959 with three other events of exportation from Japan to China (Yang Yi, ump. data).
Another possibility is that H03 was also present in Cuba, being introduced from the same native source as the Japanese population at one of the many events of introduction. It may have colonized China, later disappearing from Cuba. This hypothesis might seem less probable, but the low number of divergent sites between these haplotypes and the shared positioning on the Western clade of the phylogenetic three supports a kinship hypothesis. The native source population of bullfrogs introduced in Asian countries is probably close to the native source of the Cuban population.
Inferring the native source of invasive populations is a hard task, especially when haplotypes found in alien populations were not sampled in the native range, as observed in the present study. AMOVA results were quite inconclusive. Finer sampling of the native area with more loci sequencing is needed to elucidate the native population structure (Bai et al., 2012; Kamath et al., 2016). This way, it will be possible to better understand the origin of introduced populations. However, it is worth mentioning that our results contradict prior literature. The haplotype found by Bai et al. (2012) in China is the same we found in Asia (H03) but AMOVA indicated west USA as the native source population in their analysis, while our analysis indicated the overlap zone. We found two new haplotypes in Asia that were not sample in their work. Ficetola et al. (2008) indicates the East of USA as the native source of Belgium and Greek populations, while our results indicate the West of USA as the native source of Belgium population, after adding the new discovered haplotype.
On the other hand, we concluded that USA’s west zone is the native source of the population from Cuba and Minas Gerais, corroborating the results found by Kamath et al. (2016), as H02 is the same haplotype found by them in Grand Teton National Park (GenBank Accession number KX344492). This result adds questions about the origin of the haplotype H02, as literature indicates that founders of the population from Minas Gerais were imported from Michigan, located in the overlap zone.
Based on these analyzes, it is possible to conclude that Cuban and Southeast Asian haplotypes share native distribution, which is apart from South American’s haplotype in Eastern lineage. However, the native origin of these haplotypes cannot be designated. Nevertheless, the information gathered from the literature (Supporting Information 5) and from genetic data in the present work helps to shed light in the history of introduction of the American bullfrog (Fig. 3), being helpful for future efforts of control that might succeed by regulating sources of introduction that keep nurturing invasive populations.