Biogeography of vine thickets and open woodland in subtropical eastern Australia: a case study of three camaenid land snail genera

Sampling

We obtained genetic data from 93 specimens of land snails in the genera Brigaladra, Euryladra and Figuladra, largely collected by the first author and all now registered into the wet collection of the Queensland Museum (Supplementary Table S1). Eight additional sequences from previous phylogenetic analyses of camaenids (Clade 1, Hugall and Stanisic 2011: fig. 7; Stanisic et al. 2022) were downloaded from Genbank (Table 2). We sampled across the full geographic range of all recognised and candidate taxa. The final alignment included one outgroup sample from araucarian dry rainforest in north-eastern Queensland, Monteithosites heliostracum Stanisic, 1996. Codes used in haplotype networks and phylogenetic analyses represent nominal or putative species (identified in Stanisic et al. 2022, or proposed based on morphology and collection locality) and are as follows: AP, Figuladra appendiculata; AU, F. aureedensis; B, F. barneyae; BR, camaenid SQ16; C, F. challisi; F, camaenid BL47; FB, F. bayensis; FN, F. narelleae; IN, F. incei incei; L, F. incei curtisiana (lessoni); M, F. muirorum; MAT, Euryladra mattea; P, F. pallida and camaenid BL45; RD, F. reducta; RI, camaenid SQ15; V, Brigaladra volgiola; VL, camaenid SQ17.

Vegetation classifications

To understand how genetic diversity in Figuladra snails was distributed across habitats, we used a system of representative broad vegetation groups (BVG) that has been established for the Australian state of Queensland. Neldner and colleagues (2019b) categorise Queensland’s broad vegetation groups to reflect the vegetation structure along a mesic gradient from wet closed forests (rainforests) of the coast and north-east, to the arid spinifex hummock grasslands of the south-west. In eastern Queensland south of the Wet Tropics, the BVGs with their biogeographic and landscape attributes comprise subtropical rainforest, dry rainforest, open woodland and savanna encompassing three regions, MEQ, SEQ and SCQ (Table 1). The vast majority of snails included in the study were collected from dry rainforests (vine thickets) or open woodlands.

DNA extraction, amplification and alignment

DNA was extracted from the foot muscle using the QIAGEN DNeasy tissue kit with the Spin Column protocol. Fragments of one partial mitochondrial gene (16S) were amplified using 16SCS1 (Chiba 1999) and 16SBD1 (Sutcharit et al. 2007) primers. Each 25 μL reaction was run using 12.5 μL Phusion High-Fidelity DNA Polymerase with 0.625 μL of each primer and 0.3 μL of DNA. Reactions were performed under standard thermoprofile conditions (ThermoFisher Scientific) viz 16S mtDNA: 98°C for 2 min, followed by 35 cycles of 98°C for 30 s, 48°C for 30 s, and 72°C for 30 s and a final elongation step at 72°C for 5 min. The size of the PCR products was checked under UV transillumination after gel electrophoresis in a 1.5% agarose gel and TAE buffer solution. The PCR products were sent for purification and commercial sequencing at Macrogen Inc (Geumchum = Gu, Korea). Chromatographs were checked and aligned using the MAFFT v7.308 (Katoh and Standley 2013) plugin in Geneious 9.1.8 (Kearse et al. 2012) with default settings. We used Guidance2 (Sela et al. 2015) to identify and remove unreliably aligned regions in the 16S alignment by employing default settings.

Phylogenetic analysis

Phylogenetic analyses were performed using Maximum Likelihood (ML) and Bayesian Inference (BI). A Maximum Likelihood phylogeny was reconstructed using IQ-Tree 1.6 (Nguyen et al. 2015, univie.ac.at) under the best-fit model using ModelFinder in the software. Nodal support was inferred by performing 1000 ultrafast bootstrap replicates. Bayesian posterior probabilities were estimated by running a 1 000 000 generation Metropolis-coupled Markov Chain Monte Carlo (four chains, one heated, sampling rate 1000 generations) with a burn-in of 100 000 as implemented by MrBayes (Ronquist et al. 2012) plug-in in Geneious 9.1.8.

Population genetic summaries and demographic analysis

For each major lineage, the number of mutations, distinct haplotypes, haplotypic diversity and nucleotide diversity was determined using DnaSP v6.12.03 (Rozas et al. 2017; www.ub.edu). The number of variable and parsimony-informative sites in the alignment and pairwise sequence divergences (uncorrected-p distance) within and between target species groups were calculated using MEGA6 v10.2.6 (Tamura et al. 2013; https://www.megasoftware.net). TCS haplotype networks (Clements and Northrop 2002) for lineages were inferred using the 16S data with the program PopART (Leigh and Bryant 2015; http://popart.otago.ac.nz).

We tested for varying signatures of range expansion in each of the major lineages using two neutrality tests: Tajima’s D (Tajima 1989) and Fu’s Fs (Fu 1997) performed in DnaSP v6 using the 16S alignment data. In the absence of selection, these indices will reflect the demographic history of the populations: significant negative values are interpreted as evidence of demographic expansion and significant positive values suggest population structure. Estimated values are compared against a null distribution to test whether they are significantly different from populations where genetic drift and mutation are at equilibrium (Fig. 3).

Sequence alignment and summary details

The final alignment comprised 101 individuals, including the single outgroup sample from north-eastern Queensland (NEQ), Monteithosites heliostracum Stanisic, 1996. Modeltest, as implemented in IQ-Tree, identified TN + F + I + G4 as the best-fit model of molecular evolution. The ML alignment of all 101 sequences had 771 columns of which 267 were parsimony-informative, 52 were singleton sites and 452 were invariant sites.

Major lineages

The ML and BI analyses produced largely identical topologies (Fig. 4). In all analyses, five major strongly supported and largely geographically disjunct lineages were identified. Three of these lineages are in the genus Figuladra. The first major lineage was found in the MEQ region and is thus called the MEQ lineage. It comprises three recognised species and one putative species in mid-eastern Queensland, ranging from St Helen’s Beach in the north to Stanage Bay in the south, and adjacent islands off the coast. Species in this group occur in seasonally dry tropical forest habitats at typically low elevation, ranging from microphyll vine forests to dry araucarian rainforests. The second lineage was found in SEQ and is named as such. This lineage contains five recognised species and four putative taxa from riparian and coastal vine thickets between the Boomer Range and Inskip Point south of Fraser Island. A final lineage within Figuladra (SEQ Uplands) occurs in the foothills and low ranges of the eastern escarpments of the Great Dividing Range (GDR) and contains two recognised species occupying microphyll vine forest habitat. The range of the two species in SEQ uplands was disjunct, and seemingly interdigitated by the distribution of the SEQ clade. The MEQ and SEQ clades occur within 100 km of each other in the region corresponding to the St Lawrence Gap in mid-eastern Queensland. The other two major lineages correspond to Brigaladra from SCQ and Euryladra from SEQ and SCQ.

Fig. 4.

Phylogram of Figuladra, Euryladra and Brigaladra from eastern Queensland based on ML and BI phylogenetic analyses of 16S sequences. Numbers on top of branches represent bootstrap values, below branches posterior probabilities. Each clade is related to its associated habitat and vegetation type. Numbering and colour correspond to haplotype networks. Tip labels represent individual specimens and codes relate to species groups; these are also used in Figs 5 and 6. Scale bar indicates 3% of modelled sequence divergence.

Pairwise sequence divergences (p-distances) between the five main lineages ranged from 9.7 to 17% (Tables 3 and S2). The lowest divergences were found between the three lineages of Figuladra (9.7–13.3%); the two monotypic genera were more divergent (15.4–17%). The St Lawrence Gap marked a geographic point that separates the SEQ and MEQ lineages, and substantial genetic divergence was seen between species living on either side (11.4–13.3%) of the gap. The SEQ and SEQ upland clades are similarly divergent (9.7–13.3%), and as noted above appear to show some elevational segregation where they come into close contact, with SEQ upland being restricted to higher elevations, and SEQ more widespread in lowlands.

Genetic diversity within lineages

The better sampled and more widespread coastal and lowlands vine thicket associated SEQ and MEQ lineages of Figuladra show the highest number of distinct haplotypes, but lower values for the overall number of mutations, average pairwise differences and nucleotide diversity (Table 4, Fig. 5a, b). The MEQ lineages also show the highest values for haplotypic diversity (Hd). In contrast, seemingly disjunct populations of the SEQ upland clade from isolated patches of microphyll vine forest were sparsely sampled (n = 5) and showed the lowest number of distinct haplotypes (n = 4), but showed the highest pairwise differences (k), highest nucleotide diversity (π) and the greatest number of mutations (M) (Table 4, Fig. 5c). The widespread open woodland lineage Euryladra shows a relatively low number of distinct haplotypes, mutations and haplotypic diversity, and the lowest observed average pairwise differences and nucleotide diversity (Table 4, Fig. 6a). Finally, when compared against other lineages sampled, apparently disjunct populations of the more sparsely sampled inland vine thicket lineage Brigaladra show low numbers of distinct haplotypes and mutations, and moderate nucleotide diversity and average pairwise differences (Table 4). This lineage also shows the lowest observed values for haplotypic diversity (Table 4, Fig. 6b).

Fig. 5.

TCS haplotype network and associated maps inferred from mitochondrial 16S sequences of species associated with vine thickets. (a) MEQ lineage. (b) SEQ lineage. (c) SEQ upland with Figuladra narelleae (FN) and F. bayensis (FB). Sizes of the circles are proportional to the number of individuals with that haplotype. The number of mutational steps along the branches is indicated.

Fig. 6.

TCS haplotype network and associated maps inferred from mitochondrial 16S sequences of species associated with more open and inland woodlands. (a) Euryladra mattea (MAT). (b) Brigaladra volgiola (V) (both open woodland and semievergreen vine thicket). Sizes of the circles are proportional to the number of individuals with that haplotype. The number of mutational steps along the branches is indicated.

A negative value of Tajima’s D was inferred for the MEQ and SEQ coastal lineages as well as the wide-ranging Euryladra, suggesting recent expansion. Conversely, the two upland groups, SEQ upland and Brigaladra, show a positive Tajima’s D value, suggesting balancing selection and no recent genetic sweeps. However, for most clades these values are not statistically significant, with only the SEQ lineage having a significant negative value of the Fu’s F test (indicative of recent population expansion). Several of these lineages are considered to be complexes of several species, violating one assumption of the range expansion test (single interbreeding population) and accordingly some of these results may have to be treated with caution.

Patterns within key vegetation communities

Figuladra species from vine thicket environments on the east coast of Queensland show moderate genetic diversity with an average genetic diversity between samples of 4.6% (Table S3). Particularly high diversity with an average p-distance of 6.4%, is evident in the SEQ upland clade which mostly inhabits microphyll vine forest. The Euryladra clade from open woodlands shows lower diversities (1.9%). The upland Brigaladra, which inhabits open woodland and vine thicket on rocky scree in SCQ, shows average genetic diversity intermediate between the other two lineages (2.8%).

Upland refugia

Altitudinal endemism and biotic refugia in eastern Australia have highlighted key areas of persistence in wetter rainforests in higher montane elevations such as Eungella (Ashton et al. 2020), the Wet Tropics (Yeates and Monteith 2008) and Border Ranges (Ashton et al. 2016). Here we found that the SEQ upland lineage showed very high nucleotide diversity and p-distances, despite only five samples being available. This genetic structuring and apparently disjunct distribution suggest this lineage has a history of localised persistence in relatively upland habitats, but in an area of eastern Australia that has not, as far we are aware, previously been highlighted as a refugium. The ranges inhabited by the two species in the SEQ upland clade (100–750 m above sea level) are relatively low and characterised by araucarian microphyll vine forest, as opposed to coastal vine thicket which is often associated with slightly drier conditions (Fig. 7). Our results, and similar discoveries of potentially localised and undescribed upland endemic mygalomorph spiders (Rix et al. 2020), highlight the probability that even relatively low ranges in south-eastern Queensland may have enabled the persistence of restricted-range taxa. Other taxa characterised by localised endemism such as mygalomorph spiders remain poorly sampled in this area, emphasising a need for further surveys.

Fig. 7.

Detailed map of the distribution of the SEQ upland clade of Figuladra distribution showing specimen records for Figuladra narelleae and F. bayensis embedded within the distribution of the SEQ clade species, F. reducta.

The ranges of the two recognised species in the SEQ upland clade (F. bayensis and F. narelleae) are intersected by outwardly similar upland areas (Mt Perry and the Goodnight Scrub) that are home to Figuladra reducta, which is embedded within the more widespread SEQ clade rather than the SEQ upland lineage. There are no localities where the species within the SEQ uplands and SEQ clades occur together, with the closest region being Mt Woowoonga National Park (microphyll vine forest, F. bayensis, SEQ upland clade) and the Goodnight Scrub (dry vine thicket, F. reducta, SEQ clade) west of Gin Gin, SEQ. These two upland areas are separated by the Burnett River, a major river system in SEQ and differing vine thicket habitats. Although the exact historical processes that have shaped the distribution of SEQ upland lineages are unknown, the existence of a wide gap suggests former connectivity and then extinction in the intervening areas – perhaps driven either by climatic change, or potentially even interactions with the more widespread SEQ clade. There are no known areas of sympatry between Figuladra species, providing some support for the idea that competition or displacement may well play a role in maintaining the boundaries of populations in the SEQ and SEQ upland lineages. More geographically comprehensive and larger sample sizes coupled with genomic analyses could provide a means to further understand the embedded and contrasting distributions of the SEQ upland and SEQ lineages, with the prediction being that the former will show evidence of range contraction and bottlenecking, and the latter will show evidence of range expansion.

The Brigaladra lineage is also associated with seemingly isolated low ranges with extensive rocky habitat west of the GDR, typically, but not always, in areas of drier semievergreen vine thicket (SEVT) on scree slopes. This lineage occurs across the Broadsound, Boomer, Expedition and Peak Ranges (Fig. 8), most of which are separated by very dry lowlands dominated by open woodland in which this taxon has not been observed. Populations sampled from across the range of this taxon show genetic divergence much lower than the SEQ upland clade (average intraspecific p-distance 2.8%), and also lower than the lower elevation woodland and vine thicket clades of Figuladra. This suggests relatively recent connectivity between these inland upland areas. Profound environmental change through major glacial cycling may have influenced the distribution of the Brigaladra, and the current distribution may be relictual. This lineage also shows an ability to use open woodland environments on the boundaries of vine thickets, suggesting ecological plasticity, which may enable dispersal across apparent lowland barriers.

Fig. 8.

Detailed map of Brigaladra localities west of the GDR based on specimen records at the Queensland Museum.

Genetic patterns across the lowlands

The St Lawrence Gap is an area of relatively low and dry open woodland that is an important biogeographic barrier for many taxa, especially those associated with relatively wet habitats (Burke et al. 2013; Bryant and Krosch 2016). However, there is striking variation in both temporal and distributional divergence patterns across this region. Some wet forest taxa show very deep structuring indicative of millions of years of isolation (Hugall et al. 2003; Bryant and Krosch 2016), whereas others, especially those associated with slightly drier habitats, show shallower divergence, indicative of effectively no, or at least more recent, isolation (Bryant and Krosch 2016; Hyman and Köhler 2019). Here, we found evidence that this region is also an important turnover point for major lineages of Figuladra (MEQ and SEQ, this study) that respectively occur widely in lowland vine thickets to the north and south. Figuladra is the second lineage of large camaenid to show divergence across this gap, the other being Sphaerospira (Hugall et al. 2003). In Figuladra, this turnover occurs on both the Torilla Peninsula (Byfield NP and Stanage Bay) and adjacent areas of the mainland to the west. This suggests that this barrier of dry and seemingly unsuitable habitat was also continuous during glacial lowstands.

In contrast to the relatively deep genetic diversity and structuring in the SEQ upland clade and the St Lawrence Gap, within the two lowland clades of SEQ and MEQ we found typically low nucleotide distances between sampled populations. The haplotype network for the SEQ clade in particular tends to show only a few base pair differences between sampled haplotypes. This is also the clade that shows the strongest signature of range expansion, suggesting that through recent glacial cycles, and despite their apparently disjunct contemporary distribution, there has been quite significant gene flow between many areas of vine thicket and only recent isolation. These observations indicate that many patches of lowland vine thicket habitat have a relatively dynamic recent evolutionary history of connectivity between patches. Comparative phylogeography based on genomic datasets (i.e. SNPs) combined with demographic modelling could be undertaken to test this hypothesis further. These observations also suggest that the contemporary taxonomy of these groups, which recognise multiple species, requires further work to identify characters to support the isolation of these shallowly divergent and parapatric species.

Patterns of phylogeographic structuring in the extensive and seemingly more continuous areas of open Eucalyptus dominated open woodland in southern and central Queensland are also poorly known. The Euryladra clade is closely associated with this habitat and shows lower nucleotide diversity and average pairwise divergence than any of the other four lineages analysed, despite its wider range. One haplotype (Fig. 6a: MAT4, MAT 5, MAT10) occurs across the extreme limits of its distribution – from Targinnie near Gladstone, SEQ, and Dulacca east of Roma, SCQ, a distance of approximately 480 km. Low genetic diversity in this open woodland group indicates that there has been more extensive recent geneflow, suggesting contemporary connectivity of this habitat type (or at least at European settlement) mirrors the past distribution, with no evidence of major historical barriers.

It is widely hypothesised that fire has played an important role in shaping the distribution of open woodland and vine thicket vegetation, with the latter considered to be negatively affected by fire (Bond and Midgley 2012). Low genetic divergences between many snail populations in disjunct vine thickets suggests that the intensification of fire may have isolated many of these relatively recently in their evolutionary history. Fire also continues to pose a major contemporary threat to many vine thicket associated land snails (Stanisic and Ponder 2004), both directly and indirectly through the destruction of microhabitat. There is an urgent need to both understand biodiversity patterns in dry rainforests, and to better understand the evolutionary and ecological impacts of fire.