Evolutionary theory suggests that mono- and bispecific genera are noteworthy compared with multispecies genera because they either may be evolutionarily older (Ridley 1993) or would represent not-yet-diversified young lineages (Krug et al. 2008). In addition, many of the mono- and bispecific genera are endemic, in some cases confined to islands and archipelagoes (Ridley 1993). For example, of 83 genera of island endemic rodents (Amori et al. 2008), 57 (68.7%) were mono- and bispecific (Amori et al. 2008, 2017). The evolutionarily older phylogenesis of mono- and bispecific genera would theoretically expose them to higher rarity and extinction risk (Cotgreave and Pagel 1997). Last but not least, according to Erwin (1991), the extinction of mono- and bispecific genera is especially serious because it would mean the extinction of a full evolutionary lineage, as also demonstrated by Purvis et al. (2000) with avian and mammalian genera.

Because of their ancient history (dating back to more than 200 million yrs in the fossil record; Krenz et al. 2005), limited dispersal abilities (e.g., Zenboudji et al. 2016), and high rates of speciation in island condition (Beheregaray et al. 2004), freshwater turtles and tortoises represent ideal subjects of study for exploring the ecological correlates and the distribution features of mono- and bispecific genera as well as the worldwide distribution of their island endemics. In addition, turtles and tortoises are the most threatened vertebrates on Earth, with more than half of the species being categorized in the International Union for Conservation of Nature (IUCN) Red List as either critically endangered, endangered, or vulnerable (Stanford et al. 2018).

In this article, we 1) analyze what the distribution patterns of the mono- and bispecific genera of freshwater turtles and tortoises are, 2) analyze the distribution patterns of island endemics, 3) compare these patterns with those observed in rodents and insectivores (Amori et al. 2008, 2017), and 4) discuss some conservation issues.

METHODS

We extracted the data concerning the mono- and bispecific genera of freshwater turtles and tortoises from Rhodin et al. (2017), and their distribution and conservation status were assessed on the basis of IUCN Red List criteria (also available in Rhodin et al. 2017). For all analyses, fossil species were excluded. The biogeographical region of each taxon was classified according to Wallace (1894), revised in Rueda et al. (2013). In this regard, we used the term “Afrotropical” instead of “Ethiopian,” which was used originally by Wallace (1894).

We identified all islands/archipelagoes where at least 1 endemic species of mono- or bispecific genera of freshwater turtles and tortoises is found. An endemism index of island chelonians (%) was calculated from each island/archipelago where at least 1 endemic species occurs (data from Rhodin et al. 2017). This endemism index was calculated as the number of endemic species divided by the total number of island species.

The correlations between island/archipelago size (log transformed and expressed in km²) and 1) (log) total number of endemic species, 2) (log) endemism index, and 3) (log) total number of chelonian species were performed by the Pearson correlation coefficient. A homogeneity-of-slopes test (1-way analysis of covariance [ANCOVA]) was used to test whether the regression lines of 1) (log) island/archipelago size against (log) endemism index and 2) (log) island/archipelago size against (log) total number of species differed significantly.

Non–normally distributed variables were log transformed prior to applying any statistical analyses. Nonnormality of a given variable was assessed by Shapiro-Wilk W (assuming a nonnormal distribution when p < 0.05). We analyzed the differences in the mean number of species per genus across biogeographic regions by 1-way analysis of variance (ANOVA). Statistical differences in 1) the frequency of mono- or bispecific genera across biogeographical regions, 2) the frequency of island endemic species across biogeographical regions, and 3) the frequency of island endemic species across islands were assessed by χ² test. All statistical analyses were performed by PASW version 11.0 software, with all tests being 2-tailed and alpha set at 0.05.

RESULTS

The distribution of the number of species per genus is presented in Appendix 1. The mean number of species per genus varied significantly across biogeographic regions (1-way ANOVA: F_7,82 = 2.156, p < 0.05), and a Tukey HSD post hoc test revealed that significantly more speciose genera are found in the Oriental region.

The list of the mono- and bispecific genera of freshwater turtles and tortoises, including their biogeographic region of origin, is presented in Table 1. Overall, 28 genera were monospecific, and 18 were bispecific, accounting, respectively, for 29.8% and 19.1% of the total chelonian genera worldwide (n = 94; see Rhodin et al. 2017). In total, 48.9% of the total genera were mono- and bispecific. There was an uneven distribution of mono- and bispecific genera among biogeographical regions (χ² = 29.13, df = 5, p < 0.001), with the Oriental region housing the highest fraction of these genera, followed by the Neotropical and Afrotropical regions (Fig. 1).

Table 1. List of all the mono- and bispecific genera of freshwater turtles and tortoises, including their biogeographical region of origin.^a

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

The list of the island endemic turtles and tortoises, including their island(s) of occurrence and their biogeographical region of occurrence, is presented in Table 2. Overall, 42 species (i.e., 11.79% of the total number of species [n = 356] according to Rhodin et al. 2017) were island endemics, with most species being Neotropical (Fig. 2). If we look at the various islands as centers of endemism, the majority of the endemic island species occurred in the Galapagos Islands and in Papua New Guinea (Fig. 3). The endemism index varied remarkably among the various islands/archipelagoes, with the Philippines being the center of endemism with the minor value of the index and with Galapagos being that with the highest value of the index (Fig. 4). There was no significant correlation between (log) island/archipelago size and (log) number of endemic species (r = 0.288, p = 0.391) as well as between (log) island/archipelago size and (log) endemism index (r = 0.032, p = 0.925). However, (log) island/archipelago size was significantly positively correlated to the (log) total number of chelonian species (r = 0.709, p = 0.021) once Galapagos were removed from analyses as an outlier (if Galapagos were included, r = 0.468, p = 0.146). A 1-way ANCOVA showed that the slope of the regression line of (log) island/archipelago size against (log) endemism index differed significantly from that of the regression line of (log) island/archipelago size against (log) total number of species (F = 7.318, df = 1,19, p = 0.014).

Table 2. List of the island endemic turtles and tortoises, including their island(s) of occurrence and their biogeographical region of occurrence.^a

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

The summary of the IUCN Red List status for the monospecific, bispecific, and island endemic species of freshwater turtles and tortoises worldwide is presented in Table 3. Overall, a great portion of the monospecific (71.42%), bispecific (69.23%), and island endemics (66.6%) were threatened according to the IUCN criteria. However, there were no statistical difference in the frequency of threatened species between mono- and bispecific genera (χ² = 3.92, df = 1, p = 0.561), monospecific and island endemics (χ² = 6.43, df = 1, p = 0.376), or bispecific and island endemics (χ² = 4.32, df = 1, p = 0.633). Compared with the overall percentage of threatened taxa in freshwater turtles and tortoises (50.3%, according to Rhodin et al. 2017), species belonging to monospecific genera were significantly more frequently threatened (χ² = 77. 3, df = 1, p < 0.001), and the same was true for species belonging to bispecific genera (χ² = 59, df = 1, p < 0.001) as well as for island endemic species (χ² = 42.5, df = 1, p < 0.001).

Table 3. Summary of the IUCN Red List status for the monospecific, bispecific, and island endemic species of freshwater turtles and tortoises worldwide. Data from Rhodin et al. (2017) (see Table 2 for definition of abbreviations).

View Fullscreen

DISCUSSION

An interesting pattern emerging from our study is that the number of species per genus of turtles and tortoises varied more remarkably in the Oriental region than in all other biogeographic regions. Indeed, this biogeographic region housed both the higher number of very speciose genera (e.g., Cuora and Mauremys) and the higher number of monospecific genera (e.g., Notochelys and Orlitia), thus showing a much higher coefficient of variation in the mean number of species per genus than any other biogeographic region. We suggest that this pattern would depend merely on the higher diversity of genera occurring in the Oriental region (34% of the total extant genera) than in the other biogeographic regions (range, 3%–24%, with the Neotropical region being the second most genera rich and the Palearctic region being the least genera rich).

The main centers of island endemism for turtles and tortoises were Papua New Guinea and the Galapagos, followed by Madagascar. This result mirrors similar patterns that were uncovered also for rodents (Amori et al. 2008). In general, there was a clear and predictable area effect on the total species richness per island (i.e., the larger the island size, the higher the number of chelonian species) but not on the endemism richness per island. That is, a larger island did not house a higher number of endemic taxa than a smaller island. Because our ANCOVA revealed a significant difference in the respective slopes of these 2 relationships, it is clear that the reasons behind endemism should be divergent from the reasons behind species diversity in an island. Also, the classical island biogeography theory (MacArthur and Wilson 1967) could not be claimed to support our observed patterns given that, despite the fact that we did not measure the respective island isolations from the continental land, the area effect was very irrelevant for the endemic species. Obviously, in each island, a splitter's taxonomical view may have inflated the number of endemic taxa, or a lumper's taxonomical view may have underevaluated the number of endemic taxa. These potential biases may have in part affected the strength of our correlation analysis between island size and number of endemic species (or endemism index). Indeed, several cases are still currently debated or only recently fixed (number of species, specific or subspecific rank, generic attribution), such as within Chelydra, Macrochelys, Emys, Actinemys, Homopus, and the Galapagos giant tortoises. These taxonomically oriented biases would have been possibly serious given that the numbers of chelonian taxa per island are usually low (range, 1–13), and thus the statistical power of the analyses is clearly reduced. In addition, other controversial taxonomical cases do involve island endemics. For example, the presumed extinct Pelusios seychellensis (in theory an endemic species of the Seychelles) has been considered by several authors as simply a subspecies of Pelusios castaneus (Bour et al. 2016), thus without any endemism value for our analyses. Once more, given the small number of taxa per island, these taxonomical disagreements may have a remarkable effect on our statistical analyses and conclusions.

Overall, more than 50% of the world turtle and tortoises are threatened by IUCN criteria (Rhodin et al. 2017). Although remarkable per se, this percentage of threatened taxa was significantly less than the percentages of threatened taxa for monospecific, bispecific, and island endemics highlighted by the present study. Thus, our analyses confirmed that mono- and bispecific genera are on average more threatened than multispecies genera, as also observed in some bird and mammal groups (Purvis et al. 2000). Nonetheless, there are exceptions for this pattern. For example, the frequency distribution of the mono- and bispecific genera across IUCN categories did not differ significantly from that of multispecies genera in both rodents and soricomorphs (Solenodontidae, Soricidae, Talpidae; Amori et al. 2017).

In conclusion, our study pointed out that mono- and bispecific genera and island endemisms are particularly important for conservation policies among chelonians and that Papua New Guinea and, to a lesser degree, Madagascar should be considered priority islands for the conservation of mono- and bispecific and endemic chelonians. The Galapagos Islands also emerged as priority areas, but they already undergo a status of protected territory, and so the conservation of the endemic Galapagos tortoises is surely better achieved today than in the above-mentioned islands. Conversely, the situation is very different for Madagascar, where turtles have a status of protected species on paper (including several forest fragments that represent their natural habitats) but where there are only few and isolated actions to apply the rules in force.