Over-ocean dispersal inferred from the saltwater tolerance of lizards from Taiwan

Abstract


Abstract Background
For non-marine organisms, crossing the sea requires the challenges of dehydration and hypersalinity to be faced. Thus, the rate of water loss and saltwater tolerance determine the ability of species to disperse over sea and further in uence species distribution. However, this association between physiology and ecology has rarely been investigated in terrestrial vertebrates. In this study, we examined the lizard species differently distributed across Taiwan and the adjacent islands to determine whether these physiological responses re ect the geographical distribution. We performed immersion experiments on individuals and eggs to test the rate of water loss and the saltwater tolerance in the six chosen species, that is, four native species (Plestiodon elegans, Eutropis longicaudata, Diploderma swinhonis, and Hemidactylus frenatus) and two introduced species (E. multifasciata and Anolis sagrei).

Results
For the individuals, the results showed that P. elegans had the highest rate of water loss and the lowest saltwater tolerance, whereas E. longicaudata and E. multifasciata showed the lowest rate of water loss and the highest saltwater tolerance. D. swinhonis, H. frenatus, and A. sagrei had medium measurements compared with the aforementioned species. On the other hand, for the eggs, only the rigid-shelled eggs of H. frenatus were not in uenced by the water immersion and were incubated successfully after experimental treatment. In contrast, the parchment-shelled eggs of E. longicaudata and D. swinhonis lost or gained water dramatically in the experiments, and none of them hatched after immersion.

Conclusions
These interspecies differences in water loss and saltwater tolerance strongly suggest the heterogeneity of over-ocean dispersal ability among these lizards. Combined with the historical geology and the origin areas of each species, the inferences of the results largely explain the current distribution of these lizards across Taiwan and the adjacent islands. Furthermore, this study also helps to elucidate the potential dispersal of two invasive species, E. multifasciata and A. sagrei, for conservation purposes.

Background
Over-water dispersal makes terrestrial organisms move from one land mass to another. The capability and probability of this dispersal ecologically relate to the distribution of the creatures across lands and islands and then further evolutionarily in uence biogeography and biodiversity [1][2][3]. Molecular and fossil evidence indicates that organisms may only spread outward between lands by crossing the ocean, rather than using land bridges [2,4,5], denoting that over-water dispersal is an important strategy for organisms. In the Anthropocene, although human transportation has become a path to crossing oceans for many introduced species, for a species invading new areas arti cially, they could still spread out to adjacent areas naturally [6]. For example, Norway rats, which invaded islands around the world by arti cial boats, dispersed to adjacent islands by natural drifting or swimming [7,8]. Therefore, over-water dispersal not only is important for understanding the distribution of native species but also plays a substantial role in determining the range an invasive species can reach.
Extreme saline environments and crucial water loss are major physical challenges for terrestrial organisms when they are drifting on the ocean [9][10][11]. High-salinity environments are a stress for organisms. Different species have their own salinity tolerances [11][12][13]; this variation in uences the survival differences among species in marine oating, leading to the variance in ability of oversea dispersal at a species level that further in uences the distribution of the species [13,14]. However, research about the association between this physiological ability and distribution has mainly focused on plants, freshwater animals, and arthropods [e.g. 10,15,16]. The dispersal of terrestrial vertebrates is rarely explored. In reptiles, some freshwater and terrestrial species appear in saltwater wetlands or coasts, even surviving there for a long time [17][18][19], demonstrating that reptiles can tolerate higher salinity environments. Nevertheless, previous studies about salinity tolerance have mostly focused on marine or estuary species [18,20,21]. In the few studies on freshwater and terrestrial species, the salinity treatments were usually orally fed or injective [12,22], whereas treatment with saltwater contact, the most relevant situation faced by a drifting individual, has rarely been conducted in previous literature.
Furthermore, none of these studies link salinity tolerance to species distributions. Consequently, this key ability for crossing oceans, the saltwater tolerance of direct water contact and how it relates to distribution, has not been fully elucidated.
Dehydration is the most crucial challenge for organisms drifting on the ocean and the vital factor of saltwater tolerance. In reptiles, water loss occurs primarily through skin in addition to the loss through respiration [23,24]. The water loss of the skin is related to the adaption to the climate of the habitat environment [25][26][27], whereby species living in hot and dry habitats lose water slower than those living in cold and wet habitat [27,28]. In addition, the lower rate of water loss indicates that the species can retain water for a long time, which could delay death from dehydration [29,30]. Thus, species with lower rates of water loss are more likely to survive a drifting period of over-water dispersal than species that lose water fast. However, the relationship between water loss and survival rate in the context of ocean crossing in vertebrates has rarely been reported.
For most reptiles, there are two different ontogenetic stages to ocean crossing: the individual stage and the egg stage. Currently, ocean crossing at the individual stage has been reported abundantly through swimming, oating, or rafting [1,[31][32][33]. Compared with large species which can swim for a long distance, oating and rafting are the more likely ways for small reptiles to cross oceans [1,34]. In geckos, dispersal by natural raft or arti cial boats has been suggested in some small species [35,36]. In Anolis, it can oat for a short time on sea water [34] to disperse between islands in their native places [1,4]. On the other hand, it is probable that over-water dispersal occurs at the egg stage [35,37]. The saltwater tolerance and water loss of reptile eggs may be majorly determined by the type of eggshell. Reptilian eggs are mainly divided into two types: parchment-shelled eggs and rigid-shelled eggs [38]. The shell of the former is thinner and less calcareous than that of the latter, resulting in differences in exibility and permeability of water exchange. Parchment-shelled eggs are highly sensitive to environmental humidity [39], whereas rigid-shelled eggs, which have a dense and hard shell, could limit the exchange of water and substance from the environment [40]. Therefore, rigid-shelled eggs should have higher tolerance to seawater. Some gecko studies have shown that eggs can tolerate sea water well after immersion treatment [41,42]. For parchment-shelled eggs, successful incubation after immersion of sea water has only been reported in Anolis sagrei [43].
Taiwan and the adjacent islands are currently blocked by the sea (a depth of approximately 70 metres) from the Asian continent. However, the main island of Taiwan and the small islands between Taiwan and China were all connected to the Asian mainland before 1.55 Ma and during the Last Glacial Maximum in 26.5-18 ka, during which the sea level fell 135 metres (Fig. 1G) [44][45][46]. Therefore, Taiwan and these western islands are continental islands. Terrestrial species could disperse between Taiwan and Asian mainland through the land bridge (Fig. 1G). In contrast, the eastern ocean of Taiwan is drastically deep (a depth of over 1,000 metres) due to the oceanic trench in the nearby eastern coast, preventing the islands east of Taiwan from connecting to any island and mainland historically (Fig. 1G). Thus, the eastern islands are oceanic islands, where the species arrive only by over-water dispersal. Combining these historical contexts of geographic connection across Taiwan and these islands, knowledge of the variation of salinity tolerance among species could demonstrate how the current species distributions were formed.
In this study, we aimed to examine the heterogeneity of saltwater tolerance of small lizard species across Taiwan and the adjacent islands to determine their potential ability of over-water dispersal and then inspect the current distributions of these species and their saltwater tolerance within the context of historical geology. Six small lizard species, speci cally four native species and two introduced species, were chosen (Table 1) in this study. The four native species were Plestiodon elegans, existing throughout Taiwan, the western and northern islands, and south-eastern China (Fig. 1A); Eutropis longicaudata, distributed in southern Taiwan, the southern islands, and mainland Southeast Asia (Fig. 1B); Diploderma swinhonis, existing throughout Taiwan, the southern islands, and the Ryukyu Islands (Fig. 1C); and Hemidactylus frenatus, existing throughout Taiwan, all adjacent islands, and almost every island in East and Southeast Asia (Fig. 1D). The two introduced species were Eutropis multifasciata and Anolis sagrei. The former is native to the Philippines and was rst discovered in southern Taiwan in 1992 [47]. This species has been found in Green Island since 2008 [48] and in Orchid Island since 2017 (Fig. 1E). A. sagrei from the West Indies was discovered in central Taiwan in 2000 [49] and eastern Taiwan in 2006 [50] but has not been discovered to date in adjacent islands (Fig. 1F). In this study, we performed experiments of seawater immersion on these six species and their eggs to simulate the condition of oating on the ocean. Speci cally, we rst examined the water loss and the survival rate/incubation rate of these small lizards and their eggs using the immersion experiments to evaluate the possibility of natural cross-ocean dispersal. Second, we inspected the association between these variations in saltwater tolerance and the distributions of these six species with the historical contexts of geographic connection across Taiwan and the adjacent islands. Finally, in addition, we assessed the dispersal risk of the two introduced species for conservation purposes.
The direction is the relative position to Taiwan. The character O indicates that this species exists in that location, and the character X indicates no distribution.

Rate of water loss of lizards
Among the six species, the P. elegans individuals had the highest rate of total water loss (RTWL) in the control ( Fig. 2A), indicating that this species loses water fast in the air. In E. longicaudata and E. multifasciata, the RTWLs of the control were low, and there were no signi cant differences between the control and freshwater treatments (FW) ( Fig. 2B and F). Meanwhile, in the other four species, the RTWLs between the control and FW showed signi cant differences that were lower in FWT. In FW, except for the negative value of D. swinhonis (male: −0.069 ± 0.087, female: −0.032 ± 0.058; Fig. 2C), which indicated that D. swinhonis drank water, the RTWLs of the other ve species were not signi cantly different (male: 0.021-0.073, female: 0.019-0.063). The RTWLs were not different between males and females in all treatments in all species. Because most of the individuals in the saltwater treatments (SW) died before the end of the treatment period, resulting in the scarcity of RTWL data in this group, we did not include them in the comparison in this study.

Survival rate of lizards
The survival rates (90-100%) to the end of the treatment period were not different between the control and FW in all species (Fig. 3). In contrast, the survival rates of SW decreased signi cantly compared with the control and FW in most of the species, except E. longicaudata and E. multifasciata (Fig. 3). When treated with SW, the survival rates of the four in uenced species were extremely low, indicating their low tolerance to saltwater (Fig. 3A, C, D, and E). Meanwhile, the survival rates of the two Eutropis species were not signi cantly affected by saltwater immersion (Fig. 3B and F). Sex was not associated with the survival rate in all treatments in all species. Table 2 shows the survival in detail. P. elegans survived worse even during the treatment period where 90% of individuals in SW died on day one ( Table 2). In contrast, the survival rates of E. longicaudata and E. multifasciata remained higher than 70% at the end of the treatment. In D. swinhonis, more than 70% of SW individuals survived on the rst day; however, most of them died on the second day, and only 1 out of 20 SW individuals survived to the end of the treatment. In H. frenatus, the survival rate of SW was higher than 60% on the rst day and 40% on the second day, and 4 out of 20 individuals survived to the end. The SW survival rate of A. sagrei was higher than 70% on the rst day, and it was also 40% on the second day.
None of them survived to the end of the SW treatment.

Rate of water loss of eggs
In the eggs of E. longicaudata and D. swinhonis, the RTWLs in the control and FW were lower than zero ( Fig. 4A and B), indicating that the eggs in these two treatments absorbed water during the experimental treatments. The signi cantly lower RTWL in FW denoted that they took in much more water than the control. Meanwhile, higher RTWLs in SW than in the control and FW were shown in both species that lay parchment-shelled eggs. (Fig. 4A and B), indicating that the eggs of these two species lost water severely during SW. In contrast, the RTWLs in H. frenatus had no signi cant difference among treatments and were not different from zero, indicating that none of the treatments in uenced the water within the eggs, and almost no net water loss/absorption occurred in these rigid-shelled eggs (Fig. 4C).
Hatching rate of eggs In E. longicaudata and D. swinhonis, the hatching rates were similar between the control and FW ( Fig. 4D and E), but those in SW were signi cantly lower, indicating that saltwater immersion indeed decreased the incubation rate for these two species. In contrast, in H. frenatus the hatching rate in the control treatment (100%) was not different from that in FW (100%) and SW (100%; Fig. 4F), denoting that none of the treatments affected the hatchling rate of these rigid-shelled egg.

Discussion
This study is a rare report testing the saltwater tolerance of terrestrial vertebrates with immersion experiments. Compared with most previous studies which have focused on marine or estuary species [18,20,21,51] or those which only used orally fed or injective salinity treatments in terrestrial vertebrates [12,22,52]; in this study, we investigated both the RTWL and survival rate with saltwater immersion of these terrestrial lizards to evaluate the possibility of cross-ocean dispersal.
Saltwater tolerance of individuals among species Among the native species, our result demonstrated that the RTWL of P. elegans was the highest ( Fig. 2A) and the survival rate of SW was the lowest (Fig. 3A and Table 2), suggesting that over-ocean dispersal is less likely for this species. In contrast, E. longicaudata tolerated SW notably well, showing low RTWL in the control (Fig. 2B) and a high survival rate in SW (over 80%; Fig. 3B and Table 2), indicating that this species is more able to spread by drifting on the ocean. The other two native species, D. swinhonis and H. frenatus, had moderate tolerance to SW ( Fig. 2C and D, Fig. 3C and D, Table 2). Compared with P. elegans, they may have better ability to cross oceans if the oating period is shorter than three days ( Table 2). In introduced species, the RTWL and saltwater tolerance of E. multifasciata were similar to the native E. longicaudata ( Fig. 2F and 3F, Table 2), indicating that this invasive species had similar potential to disperse across oceans as the native Eutropis in Taiwan. In contrast, A. sagrei tolerated SW the worst; none of the individuals survived the experimental period ( Table 2), suggesting that they could not drift on the ocean for a long time. For D. swinhonis, H. frenatus, and A. sagrei, most of them died in SW; they may survive if they could cross the ocean to land within two days by drifting, which means that crossing small straits, especially among island series or archipelagos, is still possible. A. sagrei, for example, has been observed entering the sea and swimming away in its native habitat, and it could arrive to adjacent islands by oating [1]. Therefore, it is still possible that A. sagrei invades the adjacent islands from Taiwan by drifting.
The variance in survival rate in SW may be determined by the difference in RTWL among species.
Although we did not have these data in SW because of the high mortality, the RTWL values in the control and the differences between the control and FW still provides a glimpse that the amount of water loss varied among species and likely occurs in SW. Mechanically, this variance in water loss may be attributed to the differences in species body size (slower water loss in bigger species) [53][54][55], the amount of lipid content under the skin (slower water loss in species with more lipids) [25,26,54], and the morphology of scales (slower water loss in species with larger and fewer scales) [56]. We do not know the lipid contents among these species, but we did nd that the size of the scales is relatively larger in E. longicaudata and E. multifasciata than in other species (personal observation). Meanwhile, body size is not the likely factor in this study because it was not signi cant in the RTWL analysis. Evolutionarily, phylogeny [23,53] and adaptation to the habitat climate (slower water loss in species living in hotter and drier habitats) [27,28] may be associated with the variance in RTWL as well. Plestiodon originated in temperate northeast Asian [57]; thus, it is physiologically adapted to low temperatures, whereas the other species originated in tropical areas [58,59] with high temperatures. This long-term effect might lead to higher rates of water loss in P. elegans than in the other species.
Saltwater tolerance of eggs among species In the egg experiment, the eggs of E. longicaudata and D. swinhonis tolerated SW the worst, as shown in the high RTWLs and notably low hatchling rates in SW (Fig. 4A, B, D, E). In contrast, the low RTWL and the high hatchling rate in H. frenatus regardless of the treatment indicate that the eggs of this species tolerate SW well ( Fig. 4C and F), supporting the high tolerance to sea water in gecko eggs in previous studies [41,42].
The difference in SW tolerance of the eggs among these three species may have resulted from the types of egg shell; the eggs of E. longicaudata and D. swinhonis are parchment-shelled, whereas those of H. frenatus are have a rigid shell. The RTWL results in this study are consistent with previous studies showing that water permeability is usually higher in the former than in the latter [39]. We further extend this prediction from eggs in the air to eggs in the water. This variance may largely determine the egg tolerance to SW and further lead to the heterogeneity of ocean-crossing ability. Here, we suggest that successful natural drifting at the egg stage is more likely in H. frenatus than in E. longicaudata and D. swinhonis. The high SW tolerance of the gecko eggs shown in this study might partially explain the wide distribution and low genetic divergence of the populations of this species across lands and islands in Southeast Asia [59].

Interpretation of current distribution and dispersal routes in the native species
The results of this study largely re ect the current distribution of each species with the context of dispersal history and the historical topography around Taiwan. The low saltwater tolerance of P. elegans in this study explains why this species is only distributed on the western islands, Penghu and Little Liuqiu Island but not on Green Island and Orchid Island; arrival to the oceanic islands must be through overwater dispersal. P. elegans diverged with a related species that existed in Japan and the Ryukyu Islands approximately 4.23 Ma [57]. At that time, the Ryukyu Islands and Taiwan were still connected to the East Asian continent. The ancestral species may have spread southward from the Ryukyu Islands to Taiwan via the land (Fig. 5) and then spread to the western islands without crossing the ocean. Plestiodon species, belonging to the same clade as P. elegans, are almost exclusively distributed on continent and continental islands [57,60], and this supports the inference of the P. elegans dispersal route, as well.
The high saltwater tolerance of E. longicaudata demonstrates that this species has the ability to drift on the ocean for long distances and arrive on the next piece of land, explaining why this species is distributed in the oceanic islands, Green Island and Orchid Island. E. longicaudata evolved in mainland Southeast Asia approximately 38 Ma [58], possibly arriving in Taiwan via the land bridge when Taiwan was still connected to the Asian continent or through over-water dispersal by taking the oceanic current to Taiwan. Currently, E. longicaudata does not occur in northern Taiwan because of the limitation of low winter temperatures. However, if climate warming continues, it may spread northward in Taiwan and possibly cross the ocean to the north islands near Taiwan.
D. swinhonis may survive on the ocean for only one day according to our results. Therefore, ocean crossing over a long distance is unlikely for this species. The wide range of distribution they currently have (Fig. 1C) might have resulted from two routes: dispersal through the land bridge and over-water dispersal to the islands where they could arrive within one day. Because the distances between Taiwan and the adjacent islands are small (Table 1), forces from climate events, such as the monsoon or typhoons, may shorten the dispersal time to these islands to within one day. Therefore, almost every adjacent island, including the oceanic Green Island and Orchid Island, could be reached from Taiwan by this species. Their western island distributions could be explained by both land bridge dispersal and overwater dispersal, whereas the eastern island distributions resulted exclusively from the route across the ocean.
A noteworthy con ict between the results and current distribution occurs at Guishan Island. Guishan Island emerged above the ocean approximately 7,000 years ago [61] and was never connected to Taiwan by a land bridge. P. elegans, which has very low tolerance to saltwater, occurs on this island, while D.
swinhonis, which has moderate saltwater tolerance, does not. In addition, Diploderma species spreading through the Kuroshio current have occurred on the Ryukyu Islands [5] but not on Guishan Island, which is also on the route of Kuroshio. One explanation is that over-water dispersal is a rare event for both species while it only accidentally occurred in P. elegans. Alternatively, perhaps the microhabitat on this small island is suitable for P. elegans but not for D. swinhonis as these two species occupy different microhabitats; that is, the former appears in dense bush, but there are no tall trees for the arboreal D. swinhonis on Guishan Island. There is a similar situation in Penghu, which is located in the land bridge (Fig. 1G). P. elegans exists there, but D. swinhonis, which originated from China, does not [62]. The environment of Penghu may not be suitable for D. swinhonis to survive permanently because the trees are too few for this arboreal species.
H. frenatus has the widest distribution across islands, which is not in line with its moderate saltwater tolerance, but it is accordance with the excellent egg tolerance to sea water in this study. This ability may be very helpful for colonizing islands naturally and partially explain the tremendously wide distribution of this species. Geckos spreading out through areas of human activity is currently well known [36,63,64]. However, the literature usually only considers the individual stage. In this study, we showed the high potential to disperse via ocean crossing during the egg stage, matching the speculation of a previous study [35,37] and suggesting that the egg stage should be contemplated carefully in over-water dispersal in the future.

Interpretation of current distribution and dispersal routes in the introduced species
In the introduced species, there are different spreading possibilities to adjacent islands according to our results and the introduced history. E. multifasciata has been introduced arti cially, and it has spread outward from Kaohsiung in southern Taiwan since 1994 [47]. This species might reach Green Island and Orchid Island by human transportation. However, the high saltwater tolerance of E. multifasciata in our results showed that it may also be able to disperse by drifting to these islands from Taiwan or the Philippines by ocean currents. More evidence is needed for con rming how the species arrived on these islands. Today, this species has adapted well to the environment in Taiwan, and we have shown they have great ability to cross the ocean. Thus, colonizing other islands near Taiwan is highly possible for them in the future, even the northern islands if the climate warming continues.
A. sagrei are believed to have been introduced to Taiwan by human transportation, too [49]. The medium saltwater tolerance of A. sagrei in our results showed that it may drift and survive in one or two days, which is long enough to arrive to adjacent islands. Compared with E. multifasciata, A. sagrei is highly unlikely to disperse to other islands naturally because of the low saltwater tolerance and the current limited distribution in Taiwan. We suggest focusing on human activities to prevent the further invasion of this introduced species.

Conclusion
Our results showed the heterogeneity of saltwater tolerance among lizard species, in both the individual and egg stage, which may re ect the ability of over-ocean dispersal at an interspecies level. At the individual stage, P. elegans is less likely to cross oceans, whereas the two Eutropis species, E. longicaudata and E. multifasciata, are highly likely to survive on the ocean for days and successfully colonize new islands. For eggs, the rigid-shelled eggs of H. frenatus have great potential in over-water dispersal; in contrast, parchment-shelled eggs are less likely to survive after sea water immersion. Combined with the historical geology and the place of species origin, our results mainly interpret the distributions of these species across the islands around Taiwan and further propose the possible dispersal routes of these lizards. In addition, in this study, we help to elucidate the two invasive lizard species for conservation; E. multifasciata has the potential to disperse to any islands naturally, and it should be carefully monitored in future years. For A. sagrei, we should focus on the islands near Taiwan to prevent natural over-water dispersal and the islands that have ourishing agricultural transportation.

Materials And Methods
Sample collection and husbandry  019°N, 121.626°E) in Hualien. All captured individuals were kept in standard conditions for one week for acclimation before performing the lizard experiment. The pregnant E. longicaudata and D. swinhonis females were kept separately and checked frequently for eggs, which were collected while fresh. The H. frenatus eggs were collected from the wild because the pregnant females failed lay eggs in the laboratory. These eggs were used in the egg experiment. The pregnant females of viviparous E. multifasciata were also checked for hatchlings, which were used in the lizard experiment as well. Due to the high variance in body size among the captured individuals of P. elegans and the huge size difference between E. multifasiata adults and new born hatchlings, the small individuals of these two species were treated as juveniles in the lizard experiment. Only P. elegans and E. multifasiata juveniles were used in the analysis because the juveniles of the other species were di cult to collect. After the experiments, all of the lizards were released back to the places of capture. All of the collection, husbandry and subsequent treatment procedures followed the Wildlife Conservation Act of We calculated the water loss and the survival rate of the individuals and eggs to determine the effect of treatment. We de ned the day that the experiment started as the initial day and the last treatment day as the third day. After the experiments, we calculated the rate of total water loss (RTWL), where D 0 is the weight of the initial day, D 3 is the weight of the third day during the experiment, and the D w is the wastes including faeces and shedding skin, which was not used in the egg experiment.

Statistical analysis
We performed four Bayesian models to separately t the RTWLs/survival rates of the lizards and the RTWLs/hatching rates of the eggs. The RTWLs were tted by general linear models, and the survival/hatching rates were tted by logistic regressions. For tting the RTWLs/survival rates of lizard, we considered the following three factors/covariates: 1) a merging factor which merged species, sex and salinity treatments, 2) scaled mass indexes (SMIs) and 3) the interaction between the two main factors/covariates above. The SMI term was an indicator of body condition accounting for the SVL and weight of lizards and calculated according to Peig & Green [66]. Notably, we excluded the data of the SW group, and all lizards died before day three when tting the RTWLs, as most of the lizards did not survive before the end of the SW treatment. For tting the RTWLs and hatching rates of eggs, we considered the following three factors/covariates: 1) a merging factor which merged egg species and salinity treatments, 2) the weights of eggs and 3) the interaction between the two main factors/covariates above. We calculated the Bayes factor (BF) between the models mentioned above and their reduced models, where the weights of the eggs/SMI of the lizards and/or interaction terms were eliminated. Both the interaction between SMI and the sex-stage-treatment and the effect of SMI were highly unsupported in the RTWL and survival rate of the lizards (RTWL:BF < 0.01 in the interaction, BF = 0.02 in SMI; survival rate: BF < 0.01 in the interaction, BF = 0.25 in SMI); similarly, both the interaction between weight and sex-stage-treatment and the effect of weight were highly unsupported in the RTWL and hatching rate of the eggs (RTWL:BF < samples were collected after 5,000 burn-in iterations per chain. All prior samples were su ciently weak to cover the potential range of our interesting parameters, and the convergences of the MCMC samples were con rmed visually.

Declarations
Ethics approval and consent to participate All of the collection, husbandry and treatment procedures followed the Wildlife Conservation Act of Taiwan; the animal use protocols were approved by Institutional Animal Care and Use Committee of the National Museum of Natural Science (license No. 1061701832). All authors read and approved the nal manuscript.

Consent for publication
Not applicable.

Con ict of interest statement
The authors declare that they have no con icts of interest.     Dispersal routes of lizard four species in Taiwan. The orange arrow represents the routes of Plestiodon [57]. The brown represents E. longicaudata [58]. The blue green represents Diploderma [62]. The purple dotted arrow represents the possible route of E. multifasciata if it could drift to Green Island and Orchid Island. The dotted line represents the coastline during the Last Glacial Maximum. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.