Combining phylogeography with distribution modeling: multiple Pleistocene range expansions in a parthenogenetic gecko from the Australian arid zone

Abstract

Phylogenetic and geographic evidence suggest that many parthenogenetic organisms have evolved recently and have spread rapidly. These patterns play a critical role in our understanding of the relative merits of sexual versus asexual reproductive modes, yet their interpretation is often hampered by a lack of detail. Here we present a detailed phylogeographic study of a vertebrate parthenogen, the Australian gecko Heteronotia binoei, in combination with statistical and biophysical modeling of its distribution during the last glacial maximum. Parthenogenetic H. binoei occur in the Australian arid zone and have the widest range of any known vertebrate parthenogen. They are broadly sympatric with their sexual counterparts, from which they arose via hybridization. We have applied nested clade phylogeographic, effective migration, and mismatch distribution analyses to mitochondrial DNA (mtDNA) sequences obtained for 319 individuals sampled throughout the known geographic ranges of two parthenogenetic mitochondrial lineages. These analyses provide strong evidence for past range expansion events from west to east across the arid zone, and for continuing eastward range expansion. Parthenogen formation and range expansion events date to the late Pleistocene, with one lineage expanding from the northwest of its present range around 240,000 years ago and the second lineage expanding from the far west around 70,000 years ago. Statistical and biophysical distribution models support these inferences of recent range expansion, with suitable climatic conditions during the last glacial maximum most likely limited to parts of the arid zone north and west of much of the current ranges of these lineages. Combination of phylogeographic analyses and distribution modeling allowed considerably stronger inferences of the history of this complex than either would in isolation, illustrating the power of combining complementary analytical approaches.

Figure 1. Haplotype network for the 3N1 mtDNA lineage, showing nesting levels.

Clades correspond to those listed in Table 2. Small ovals without letter names are haplotypes not sampled but which are necessary to connect sampled haplotypes. Pie charts next to each haplotype indicate the proportion of individuals with that haplotype sampled from the various regions described in the analytical methods and Figure 7.

Figure 2. Haplotype network for the 3N2 mtDNA lineage, showing nesting levels.

Clades correspond to those listed in Table 2. Small ovals without letter names are haplotypes not sampled but which are necessary to connect sampled haplotypes. Pie charts next to each haplotype indicate the proportion of individuals with that haplotype sampled from the various regions described in the analytical methods and Figure 7.

Figure 3. Mismatch distributions for 3N1 and 3N2 lineages.

Distributions are shown for all 3N1 or 3N2 populations together and for eastern and western 3N1 populations.

Figure 4. Statistical distribution models for parthenogenetic Heteronotia binoei.

Models are for (a) present conditions and (b–d) last glacial maximum conditions with a 9°C reduction in mean air temperature and three different rainfall reduction scenarios. All statistical models are based on the AICc model reported in Kearney et al (2003). Overlayed on the predicted distributions are the contours for biophysical predictions of the limit for successful development of eggs (600 degree days), of the limit for potential activity time (0 hours) and of the number of hours of potential activity at the current distribution limit of the Heteronotia complex (400 hours). Any regions roughly south of the contours are outside the fundamental niche of Heteronotia. The biophysical predictions use either (a) current climatic conditions or (b–d) a 9°C decrease in monthly maximum and minimum temperatures. Methods for biophysical predictions are described in Kearney and Porter (2004).

Figure 5. Proposed origin and spread of 3N1 and 3N2 parthenogens.

Also shown are timing estimates for expansions and hypothesized future expansions in 3N1 parthenogens. Phylogeographic events are overlaid on the predicted distribution for parthenogenetic Heteronotia binoei based on a statistical distribution model for present climatic conditions . Times given here are point estimates; confidence intervals are given in Table 2. DRD = dispersal restricted by distance.

Figure 6. Sampling localities for the 3N1 and 3N2 mitochondrial clones.

Latitude/longitude data and sample sizes are given in Table 5. Ranges of the CA6, EA6, and SM6 sexual chromosome races are shown in light gray. The inset in the upper right shows the extent of the arid zone in dark gray.

Figure 7. Regions used for various analyses.

Ranges for each mtDNA lineage as a whole are shown in light gray.