Comparative Phylogeography of Freshwater Habitats in Southern Australia: Using Palaeodrainage Reconstructions to Investigate Population Structure and Historic Population Connectivity
In:
Submitted By sm3jackson Words 6481 Pages 26
Comparative Phylogeography of Freshwater Habitats in Southern Australia: Using Palaeodrainage Reconstructions to Investigate Population Structure and Historic Population Connectivity
Sarah Jackson
A thesis submitted in partial fulfilment for the degree of Bachelor of Animal and Veterinary Bioscience with Honours
Department of Genetics School of Molecular Science La Trobe University
October, 2011
Contents
Abstract ........................................................................................................................... iii 1. Introduction .................................................................................................................. 1 1.1 Phylogeography....................................................................................................... 1 1.2 Palaeodrainage........................................................................................................ 4 1.3 Background on Study Species .................................................................................. 6 1.4 Aims and Hypothesis ............................................................................................... 8 2. Materials and Methods ............................................................................................... 10 2.1 Locations ............................................................................................................... 10 2.2 Expectations of Hypothesis.................................................................................... 12 2.3 Sample Collection and Storage .............................................................................. 13 2.4 DNA Extraction ...................................................................................................... 13 2.5 Pilot study ............................................................................................................. 13 2.6 Amplification ......................................................................................................... 15 2.7 Purification and Sequencing .................................................................................. 15 2.8 Sequence Alignment and Analysis ......................................................................... 16 3. Results ........................................................................................................................ 19 3.1 Paratya australiensis.............................................................................................. 19 3.1.1Mitochondrial Data ......................................................................................... 19 3.1.2 Phylogenetic Analysis ..................................................................................... 19 3.1.3 Haplotype Network ........................................................................................ 24 3.1.4 Summary Statistics ......................................................................................... 26 3.2 Austrochiltonia subtenuis ...................................................................................... 28 3.2.1 Mitochondrial Data ......................................................................................... 28 3.2.2 Phylogenetic Analysis ..................................................................................... 28 3.2.3 Haplotype Network ........................................................................................ 32 3.2.4 Summary Statistics ......................................................................................... 34 4. Discussion ................................................................................................................... 35 4.1 Original Hypothesis ............................................................................................... 35 i
4.2 Phylogeographic Pattern ....................................................................................... 35 4.3 Non-Biogeographic Explanation for Pattern Observed ........................................... 39 4.4 Paratya australiensis, One Species or Many? ......................................................... 39 4.5 Future Research .................................................................................................... 40 4.6 Conclusions ........................................................................................................... 42 Acknowledgements......................................................................................................... 44 References ...................................................................................................................... 45 Appendix 1: Stock Solutions Used in Materials and Methods .......................................... 52 Appendix 2: Sample Site Locations .................................................................................. 53 Appendix 3: McCluskey Sequence Information ............................................................... 54 Appendix 4: King and Leys Sequence Information ........................................................... 57 Appendix 5: Paratya haplotypes ..................................................................................... 58 Appendix 6: Austrochiltonia haplotypes .......................................................................... 59
ii
Abstract
The rise and fall of sea levels due to glaciations has resulted in continual change of the coast line of Australia, the most notable being the formation of a land bridge between Victoria and Tasmania. Low sea levels expose palaeodrainage networks connecting currently isolated drainage basins and allowing gene flow between populations, while rising sea levels submerge networks and disconnect these populations. These palaeodrainage networks are believed to have played a part in the distribution of freshwater dependant species. In this study mitochondrial COI sequences were examined from approximately 100 freshwater shrimp (Paratya australiensis) and 68 freshwater amphipods (Austrochiltonia subtenuis) from around Port Phillip Bay and across the western coast of Victoria, with the expectation of finding evidence of historic population connectivity coinciding with palaeodrainage networks. Phylogenetic and phylogeographic patterns and demographic stability measures were analysed. Surprisingly, very little phylogeographic structure was seen in either species, with single haplotypes shared across many sites over broad geographic ranges independent of both palaeodrainage and present day drainage connections. This broad drainage independent dispersal obscured the potential to detect gene flow across palaeodrainage networks. However, these results are significant in themselves as they are far from what would be expected of freshwater dependent macroinvertebrates.
iii
1. Introduction
1.1 Phylogeography
Having a good understanding of the biogeographic history of populations can have vast impacts on the way we think about population structure, assisting in the maintenance of genetically viable populations by identifying and prioritising areas for monitoring, management and protection (Moritz and Faith 1998). Phylogeography is the study of biogeographic history: it combines the current knowledge on species biology, phylogenetic structure and geography (both past and present) (Beheregaray 2008). It involves the construction of gene genealogies that are then superimposed over geographic locations to determine the biogeographic history of populations (Zink 2002). Since Avise et al. (1987) first described the concept of phylogeography, the field has been growing rapidly, with the number of publications including the term increasing exponentially (Beheregaray 2008, Kholodova 2009).
Phylogeographers are able to test biogeographic hypotheses, assess the formation of cryptic species and determine the origin, distribution and maintenance of biodiversity (Beheregaray 2008). Inferences about past geological conditions can also be made by assessing the structure of population genealogies (Zink 2002). Because of this, phylogeography has been able to make significant contributions to areas of study such as speciation, historical biogeography, human evolution, conservation biology, biodiversity research, taxonomy, palaeoecology, palaeoclimatology and vulcanology (Beheregaray 2008).
1
Comparative phylogeographic studies investigate multiple taxa sharing similar geographic distributions. Intraspecific comparisons may identify geographic areas where whole communities have undergone independent evolution (Moritz and Faith 1998). Comparative methods can have a big influence on the identification of areas of conservation as they extend the application of phylogeography to identifying whole ecosystems of significance rather than just populations of single species (Moritz and Faith 1998, Zink 2002, Kholodova 2009).
For a freshwater dependent species the terrestrial environment and ocean both represent formidable barriers to dispersal, making them highly reliant on freshwater connectivity for dispersal and gene flow (Chaput-Bardy et al. 2008). This makes them of particular interest in phylogeographic studies as the distribution of their lineages is often tightly linked to the landscape (Bermingham and Avise 1986, Avise 2000, Unmack 2001). Numerous studies of freshwater species have demonstrated high levels of gene flow within catchments but little between them (Carini and Hughes 2004, Baker et al. 2004b, Huey et al. 2006). A clear example of this is in the freshwater eel-tailed catfish Neosilurus hyrtlii and Porochilus argenteus (Huey et al. 2006). This study showed high levels of gene flow within catchments but no gene flow between neighbouring catchments. In contrast, aquatic species with a terrestrial life stage are often more likely to exhibit gene flow across much broader areas due to greater dispersal abilities. For example, Artiss (2004) found little genetic structure and high levels of gene flow among catchments and drainages in the dragonfly Libellula quadrimaculata across Asia, Europe or North America. Similarly in a European study on the mayfly Baetis rhodani, Williams et al. (2006) also
2
observed a widespread distribution across Europe of this species with no obvious geographic pattern.
Despite the obvious constraints on freshwater species with no active dispersal stage, many still have widespread distributions (Bohonak and Jenkins 2003). Many species such as the freshwater snail Radix balthica (Pfenninger et al. 2011) and fish such as the spangled perch Leiopotherapon unicolor (Bostock et al. 2006) appear to have achieved their widespread distributions by passive dispersal through the use of mechanisms such as animal vectors or wind (Bilton et al. 2001). Alternatively freshwater-limited species such as the eel-tailed catfish discussed earlier (Huey et al. 2006) have relied on changes in connectivity among drainages to gain wide distributions. There are a number of circumstances where river systems may have connected historically and facilitated dispersal between currently isolated populations or fragmentation of currently connected populations (Craw et al. 2007, Burridge et al. 2008, Hughes et al. 2009) (Fig. 1).
Geographic changes such as tectonic events, natural damming and erosion can cause the ‘capture’ of a river tributary by a neighbouring drainage, resulting in previously connected populations becoming isolated and previously isolated populations becoming connected (Hughes et al. 2009) (Fig. 1A). The changing sea levels over the past (80,000 years) (Lambeck and Chappell 2001) may cause coastal branching streams to be below sea level and become isolated from each other. In this case dispersal can only occur during times of low sea level via palaeodrainages (Schultz et al. 2008) (Fig. 1B). Dispersal may also occur during flooding across a low lying divide between two drainages (Fig. 1C). In cases where drainage basins are divided in low relief valleys, erosion and sediment 3
build up can lead to changes in flow direction; these episodic connections allow individuals to cross from one basin to another (Fig. 1D). Lakes forming at a divide between two drainages may connect to either or both drainage basins (Fig. 1E), and finally swampy ground between two drainage basins may provide a suitable environment for freshwater dependent species to cross drainages (Fig. 1F).
Figure 1 From Burridge et al. (2008). Six possible connections that may form between adjacent drainage basins that allows dispersal.
1.2 Palaeodrainage
When a river changes its course or is cut off by rising sea levels, the bed where the river historically flowed is referred to as a palaeodrainage basin (Schultz et al. 2008). Throughout the late Pleistocene the sea level change due to the exchange of water between glaciers and oceans occurred frequently. During ice ages sea levels are low providing more connections between freshwater drainages and during interglacial 4
periods water released from glaciers causes sea levels to rise and previously connected drainages are separated (Lambeck and Chappell 2001) .
The influence of lowered sea levels and the connection of freshwater populations via palaeodrainage have been seen around the world (Shih et al. 2006, Craw et al. 2007, de Bruyn and Mather 2007). For example in the freshwater crab Somanniathelphusa taiwanensis gene flow is expected to have occurred between China and Taiwan via a land bridge during the last glaciation (Shih et al. 2006). In the north of Australia dispersal via palaeodrainage was found in the freshwater shrimp Macrobrachium rosenbergii across the Indo-Australian Archipelago (de Bruyn and Mather 2007).
Sea levels have been low enough to provide a land bridge between Victoria and Tasmania numerous times in the last 80,000 years (Lambeck and Chappell 2001) (Fig. 2). Due to these lowered sea levels, it is hypothesised that freshwater species along the southern coast of Victoria and the northern coast of Tasmania have been directly influenced by palaeodrainage connectivity (Fig. 3) providing multiple opportunities for dispersal and gene flow among presently isolated populations and providing a direct biogeographic link between Victoria and Tasmania (Horwitz 1988).
Figure 2 From Lambeck and Chappell (2001). Historical reconstruction of the Australian Bass Strait area coast line during the last glacial maxima. 5
Two recent studies have demonstrated the influence of palaeodrainage connections on freshwater species across Victoria and Tasmania. Shultz et al. (2007) observed close relationships in a freshwater crayfish species Geocharax gracilis, between drainages in Victoria and King Island in Bass Straight. Similarly, Coleman et al. (2010) found that relationships between Victoria and Tasmania freshwater dwarf galaxias Galaxiella pusilla populations, were best explained by past palaeodrainage connections. Given that these studies of two very different species show similar patterns of palaeodrainage-related biogeography across presently unconnected freshwater systems it seems likely that the lowering and raising of sea levels over the past 80,000 years is directly responsible for shaping freshwater biodiversity in this region.
1.3 Background on Study Species
An organism used in a biogeographic study on freshwater drainage basin history should ideally have relatively low inter-drainage dispersal ability and a fully aquatic life history. This is to ensure genetic relationships observed are due primarily to river connectivity rather than terrestrial dispersal (Bunn and Hughes 1997). For these reasons both the freshwater shrimp Paratya australiensis and amphipod Austrochiltonia subtenuis were selected for this study.
P. australiensis has a broad distribution, being found inhabiting coastal and inland rivers, streams, lakes, and farm dams. It is distributed from far north east Queensland through to South Australia (Williams 1977), it is also found throughout Tasmania. It is however believed to be a complex of cryptic species with studies being conducted throughout its
6
range in Queensland revealing a number of very divergent lineages (Cook et al. 2006). P. australiensis is predominantly believed to be intolerant to saline environments being a part of the freshwater family Atyidae (Williams 1977, Walsh and Mitchell 1995). However there have been studies that have found them in estuarine environments along the southern coast of Australia (Walsh and Mitchell 1995). P. australiensis give birth to planktonic larvae with a duration of up to 45 days (Walsh 1993), which may aid in its dispersal (Baker et al. 2004b).
A. subtenuis is also found distributed widely across southern Australia, including Tasmania. It is found most commonly in the western part of Victoria and is rarer across the eastern part (Lim and Williams 1971). A. subtenuis has no planktonic larval stage:
instead females carry fertilized eggs with them (Lim and Williams 1971). As such their dispersal abilities are expected to differ from those of P. australiensis. In a lab based study by Lim and Williams (1971) on the desiccation rate of A. subtenuis, they were found to survive no longer then 30 minutes out of water. This, combined with the knowledge that Austrochiltonia in other areas of Australia presented as a complex of cryptic species (Murphy et al. 2009), leads to the expectation that similar high levels of genetic structure will be seen across the coastal rivers of Victoria.
The dispersal ability of P. australiensis is expected to be greater than that of A. subtenuis as it is larger and is free swimming and has an active larval stage which will be subject to the impacts of river flow, while A. subtenuis is not an active swimmer and is found more to crawl. It is therefore expected that A. subtenuis will be more constrained by the
7
freshwater drainages and show higher levels of phylogenetic structure both within and between drainages.
1.4 Aims and Hypothesis
This study has two main hypotheses. Firstly, it is hypothesised that the freshwater limited life history of P. australiensis and more particularly A. subtenuis should result in a strong phylogeographic structure between isolated drainages as seen in other freshwater shrimp (Hurwood et al. 2003, Baker et al. 2004b) and amphipods (Murphy et al. 2009). Secondly, as palaeodrainage connections recently as 20,000 years ago (Lambeck and Chappell 2001) appear to have played a significant role in the distribution of freshwater species inhabiting coastal river systems in Victoria (Schultz et al. 2008, Coleman et al. 2010), it is hypothesised that there should be less genetic structure across populations historically connected by palaeodrainages and higher levels of genetic structure across populations that have never been connected (Fig. 3).
If dispersal has not occurred through palaeodrainages as hypothesised then there will be no phylogeographic signal evident in previously connected drainages. Given the scarcity of life history information on the majority of Australian freshwater invertebrates, it is also possible that the dispersal limitation on the two target species has been overestimated and there may be a lack of phylogeographic structure among any of the drainages. However, given previous studies of Austrochiltonia amphipods have demonstrated that significant phylogeographic structure is found at very small scales (Murphy et al. 2009), this is considered unlikely.
8
Figure 3 Adapted from map provided by Daniel Ierodiaconou of Deakin University, Warrnambool. Historic palaeodrainage basins when sea levels were 120 meters below present. The Port Phillip Bay palaeodrainage is shown in purple, the Cape Otway palaeodrainage is shown in green, the Gippsland palaeodrainage is shown in yellow, and the Hopkins and Portland Coast drainages show in blue and orange. Historically Tasmania was joined to the mainland by palaeodrainage connections through Bass Lake. Based on the hypothesis, rivers that have been connected by palaeodrainage (i.e. those that are the same colour) should show less phylogeographic structure compared to rivers that have never been connected (i.e. among different coloured rivers).
9
2. Materials and Methods
All solutions referred to throughout this section are listed in Appendix 1.
2.1 Locations
A common method for determining the position of past shorelines and mapping out palaeodrainage networks is the use of bathymetric depth contours, where the seafloor depth is used to map past shorelines at varying water depths (Lambeck and Chappell 2001, Schultz et al. 2008). Using this method palaeodrainage reconstructions of Bass Strait used in this study were provided by Daniel Ierodiaconou of Deakin University, Warrnambool.
Based on this palaeodrainage reconstruction a number of sites across Victoria were selected to test the hypothesis (Table 1, Fig. 4). Sites selected included Barwon River (BAR), Lerderderg River (LER), Jacksons Creek (JAC), Deep Creek (DEP), Merri Creek (MER), Plenty River (PLE), Watsons Creek (WAT) and Bunyip Rive (BUN), which all connected via the main ‘Port Phillip Bay drainage’ (Fig. 4; Purple). Samples were also collected from Curdies River (CUR), Gellibrand River (GEL) and Aire River (AIR), which were all connected via the ‘Otway Coast drainage’ (Fig. 4; Green), Franklin River (FRA), which is a part of the ‘Gippsland drainage’ (Fig. 4; Yellow) and from Moyne River (MOY) which has never been connected via palaeodrainage (Fig. 4; Orange). To compare finding with dispersal across current day drainage connections, samples were collected from two separate rivers in the Hopkins Catchment: Mt Emu Creek (MtE) and Hopkins River (HOP) (Fig. 4; Blue) and a site downstream of the confluence of these rivers: Hopkins Falls (HOPF), these also formed a separate palaeodrainage. Paratya from Coal River on the south east of Tasmania (not shown in Fig. 4) were also included. 10
Figure 4 Adapted from map provided by Daniel Ierodiaconou of Deakin University, Warrnambool. Drainage basins based on present day and historic sea level. (Left) Present day drainage basins (Right) Historic palaeodrainage basins when sea level 120m below present. See table 1 for definition of site abbreviations.
11
Table 1 Sample locations, drainage basins and the number of Paratya (nP) or Austrochiltonia (nA) and haplotypes found in these locations in each location (hapP and hapA. In locations marked with * only Austrochiltonia australis was found. For full haplotype descriptions refer to Appendix 5 and 6.
Site Abv. FRA BUN WAT MER PLE JAC DEP LER BAR AIR GEL1 GEL2 CUR MtE HOP HOPF MOY CAM COL
Location Franklin River Bunyip Watson Creek Merri Creek Plenty River* Jacksons Creek Deep Creek Lerderderg Barwon River Aire River Gellibrand River Gellibrand River Curdies River Mount Emu Creek Hopkins Hopkins Falls Moyne River Campaspe River* Coal River TAS
Drainage South Gippsland Bunyip river Yarra river Yarra river Yarra river Maribyrnong river Maribyrnong river Werribee river Barwon river Otway Coast Otway Coast Otway Coast Otway Coast Hopkins River Hopkins River Hopkins River Portland Coast Campaspe river Coal River
2.2 Expectations of Hypothesis
Based on the hypothesis and palaeodrainage reconstructions it is expected that little phylogeographic structure will be evident within the sites in the present day Hopkins catchment (blue) as well as within the historically connected Otway Coast palaeodrainage (green) and the within Port Philip Bay palaeodrainage (purple). Conversely a high degree of genetic structure is expected between the Hopkins, Otway Coast and Port Philip drainages, as well as Moyne River (Orange) and Franklin River (Yellow) as there has never been a palaeodrainage connection between these sites. 12
2.3 Sample Collection and Storage
Samples were collected with sweep nets from rivers across the western coast of Victoria as shown in figure 4 and table 1. Where possible samples of A. subtenuis and P. australiensis were both collected, however both species were not present at every site (Table 1). Due to morphological similarities to A. subtenuis, A. australis from Campaspe River and Plenty River (Table 1) was inadvertently collected and sequenced. Positional data was recorded at all sites using a hand held GPS (Appendix 2). All samples were stored in 95% ethanol and kept in an esky and were stored in the fridge on return to the lab. Samples were stored in 95% ethanol below 4°C until DNA extraction.
2.4 DNA Extraction
The Chelex DNA extraction protocol (adapted by Nick Murphy) was used. Single legs from each shrimp and entire amphipods were used in extractions. Each extraction contained 250µl of 5% Chelex/TLE, tissue sample and 2µl of 20mg/ml Proteinase K. Extractions were kept at 55°C overnight to digest tissue. Extractions were then heated to 95-100°C for 10 minutes, and then allowed to return to room temperature. Each extraction was then vortexed for 1 minute and centrifuged at 13,000rpm for 1 minute. A 1/10 dilution working solution was made for each extraction using 5µl supernatant and 45µl TLE. The working solution stored at 4°C and the DNA extraction stored in freezer at -20°C.
2.5 Pilot study
As both species being used have not been studied often pilot studies were required to establish effective primers and PCR conditions to amplify the required gene. A fragment of the mitochondrial cytochrome oxidase I (COI) gene was chosen for this study as it is 13
used widely as a DNA ‘barcoding’ gene for species identification and evolves rapidly enough to provide insight into deep phylogeographic relationships (Herbert et al. 2003).
For the shrimp P. australiensis the general invertebrate COI primers HCO1 and LCO1 (Folmer et al. 1994) and C1-J-1718 and C1-N-2329 (Simon et al. 1994) were both tested at temperature gradients from 40°C to 50°C. HCO1 and LCO1 did not work consistently for all DNA extracts and produced faint products with double bands when run on Agarose gel. C1-J-1718 and C1-N-2329 successfully amplified the target gene in most extracts and produced clear PCR products with no double bands at 50°C. As PCR product was faint PCR cycles were increased to 35 and an extra 1.5µl Mg2+ was added as it can aid in binding of primers to target site.
For the amphipod A. subtenuis the general invertebrate COI primers HCO1 and LCO1 (Folmer et al. 1994) and C1-J-1718 and C1-N-2329 (Simon et al. 1994) were also tested at temperature gradients from 40-50°C. HCO1 and LCO1 produced very faint products and only worked on a couple of samples tested, the addition of Mg2+ did not improve the reaction. The primers C1-J-1718 and C1-N-2329 did not work under any condition tested. The amphipod specific primers M202 and M493 (King and Leys 2011) and M735 and 737 (Murphy et al. 2009) all only worked sporadically. Therefore primers more specific of A. subtenuis were designed using A. subtenuis sequences from GenBank (NM-35, TAGGTGCTTGGGCGAGAGCTGTAG and NM-36, TAGCAGTAATAAACACAGACC) which amplified an approximately 500bp section of COI. NM-35 and NM-36 worked for most samples. However due to double banding at lower temperatures PCR’s were run with an annealing temperature of 60°C. 14
2.6 Amplification
A ~500bp fragment of the protein coding mitochondrial COI gene was amplified for A. subtenuis using the primers NM-35 and NM-36 as mentioned above. Amplifications were performed using the polymerase chain reaction (PCR) in a 25µl total volume containing: 14.5µl distilled PCR grade water, 2.5µl HS buffer, 2µl 10mM dNTP, 2µl of each primer, 0.1µl HS Taq and 2µl DNA template. A blank sample was used in each reaction. Thermal cycling was performed using the BIO-RAD S1000 Thermal Cycler with the initial denaturing step of 95°C for 2 minutes, then 33 cycles of 95°C for 20 seconds, 60°C for 30 seconds, 72°C for 30 seconds and a final extension step of 72°C for 5 minutes.
A ~600bp fragment of the COI gene was amplified for P. australiensis using the primers C1-J-1718 and C1-N-2191 as mentioned above. Amplifications were performed using PCR in a 25µl total volume containing: 13µl distilled PCR grade water, 2.5µl HS buffer, 2 µl 10mM dNTP, 2µl of each primer, 1.5µl Mg2+, 0.1µl HS Taq and 2µl DNA template. A blank sample was used in each reaction. Thermal cycling was performed using the BIO-RAD S1000 Thermal Cycler with the initial denaturing step of 95°C for 2 minutes, then 35 cycles of 95°C for 20 seconds, 50°C for 30 seconds, 72°C for 30 seconds and a final extension step of 72°C for 5 minutes.
2.7 Purification and Sequencing
2µl of each PCR product was mixed with small drop of Bioline 5x DNA loading buffer and electrophoresed on 2% agarose gel for 20 minutes at 100 volts. Gels were then stained for 30 minutes in a bath of Gel Red Stain on a rocking platform and visualised using the BIO-RAD Gel doc TR+ to determine the success of the PCR. Unbound primers and dNTPs 15
were removed from PCR products using the protocol for the MultiScreen PCR plate. Purified PCR products were again electrophoresed and viewed using the BIO-RAD Gel doc TR+ to determine volume of PCR product required for sequencing (judged by the intensity of the bands).
Sequencing reactions were performed using the Australian Genome Research Facility (AGRF) protocol for Big Dye Terminator labelling reactions. Thermal cycling was again done using the BIO-RAD S1000 Thermal Cycler with the initial step of 96°C for 2minutes, then 27 cycles of 95°C for 10 seconds, 50°C for 5 seconds, 60°C for 4 minutes and the final step of 4°C for 5 seconds. Sequencing reactions were cleaned up using the protocol for the Millipore MultiScreen SEQ plate. Purified samples eluted in 20µl PCR grade H2O were sent to the Australian Genome Research Facility (AGRF) (www.agrf.org.au/) for sequencing. Any ambiguous sequences were re-sequenced using the reverse primer.
2.8 Sequence Alignment and Analysis
All sequences were checked for contamination and species confirmed by doing a comparison with sequences in GenBank using the Basic Local Alignment Search Tool (BLAST) software (www.ncbi.nlm.nih.gov/blast/). COI sequences were analysed and aligned in Geneious 5.4.4 (Drummond et al. 2011) using a Geneious Alignment. Alignments were confirmed by eye, low quality regions at the end of each sequence were trimmed and ambiguous bases replaced with Ns.
To broaden the number and range of samples used the Paratya dataset was combined with 32 COI haplotypes from 71 locations across Victoria and south east South Australia 16
downloaded off GenBank from the PhD thesis by McCluskey (2007) (Appendix 3) and the outgroup used to root trees was a Paratya howensis sequence (McClusky 2007). The Austrochiltonia data set was combined with 61 COI sequences from 23 locations from the central and Port Phillip areas of Victoria and south east of South Australia downloaded off GenBank from King and Leys (2011) (Appendix 4). Of these sequences 23 were found to belong to A. subtenuis and 6 to the closely related and morphologically similar A. australis. The outgroup used to root trees was a Hyalella azteca sequence from GenBank (Baird et al. 2011).
The appropriate model of evolution for each alignment was determined using jModelTest 0.1 (Posada 2008), using the hierarchical likelihood ratio test (hLRT) and Akaike Information Criterion (AIC). The optimum evolutionary model for Paratya was the GTR+I+G, with a gamma shape distribution parameter of 0.143, and for the Austrochiltonia alignment the optimum evolutionary model was GTR+I+G, with proportion of invariable sites of 0.371, and a gamma shape distribution parameter of 0.501.
Maximum Likelihood (ML) and Bayesian analysis methods were used to generate phylogenetic trees to assess the genetic relationships between samples. ML analysis was performed using PHYML (Guindon et al. 2010) with 1,000 bootstrap replicates generated. Bayesian analysis was performed using MrBayes 3.1.2 (Huelsenbeck and Bollback 2001). Analyses were run across two chains for ten million generations sampling every hundred generations. Stationarity was determined by examining the log likelihood generated by the program. A burn in of 10,000 generations was set to discard trees created prior to 17
stationarity, and Bayesian posterior probabilities were calculated for remaining trees to represent the percentage of times each node was recovered. Genetic divergence between and within lineages observed in ML and Bayesian analysis were calculated using Kimura 2 Parameter genetic distances in Mega 5.05 (Tamura et al. 2011).
Median Joining networks (Bandelt et al. 1999) were constructed using the program Network 4.6 (http://fluxus-engineering.com) to determine if there is any significant relationship between haplotypes observed and geographic locations. As the program Network requires all sequences to be equal lengths Ns were added to the end of short sequences to the required length. As samples collected in this study have only a limited alignment with those downloaded from GenBank (~300bp) the combined data set was not used in the construction of Networks. The full A. subtenuis dataset collected in this study was used to construct a Network. As a number of highly divergent lineages were revealed by phylogenetic analysis of P. australiensis, the network only included samples from the main lineage B as it was found across western Victoria and Port Phillip Bay in the rivers historically connected by palaeodrainage.
To compare genetic diversity within and between populations, haplotype diversity (h) and nucleotide diversity (π) were calculated. Tajima’s D (Tajima 1989), Fu and Li’s D* and F* (Fu and Li 1993), Fu’s Fs (Fu 1997) and the R2 statistic (Ramos-Onsins and Rozas 2002) were used to assess demographic stability, to see if there is there is evidence of selection, population growth or population bottlenecks affecting populations. These statistics were all calculated using the program DnaSP v5 (Librado and Rozas 2009) and the significance of each demographic statistic was calculated using 1000 random coalescent simulations. 18
3. Results
3.1 Paratya australiensis
3.1.1Mitochondrial Data A total of 110 individual shrimp from 15 locations around Port Phillip Bay and the western coast of Victoria (Table 1; Fig. 4) were sequenced for an approximately 600 base pair fragment of the mtDNA gene COI. Out of the 582 aligned nucleotide sites there were no indels. A total of 27 unique haplotypes were found (Table 1).
3.1.2 Phylogenetic Analysis Phylogenetic reconstructions based on Maximum Likelihood (ML) and Bayesian analyses resulted in very similar tree topologies, with support values produced by each analysis being comparable. The Bayesian tree is shown including both Bayesian posterior probability values above 0.95 and ML bootstrap values above 70 shown at the branches (Figure 5), as these values demonstrate significant confidence in a branch. Four distinct lineages were observed, and were labelled lineage A, B, C and D in correlation with lineages observed by McCluskey (2007). High levels of genetic divergence was seen between these lineages with between 6% and 10% genetic distance compared to between 0% and 3% genetic distance within lineages (Table 2).
Lineage B is the most frequently occurring clade both in samples collected in this study and by McCluskey. The distribution of this lineage covered the rivers spanning the west coast of Victoria and Port Phillip Bay including rivers hypothesised to be connected by historic palaeodrainage connections (Fig. 6B). It is also found across rivers in South Gippsland on the east coast of Victoria, Loddon, Campaspe, Wimmera and Ovens basins 19
in inland central Victoria, the Glenelg drainage basin on the far west coast of Victoria, the Murray River drainage basin in South Australia and the Coal River, Tamar River, East Coast, and Hently River drainage basins in Tasmania. Each of these drainage basins (apart from Ovens) is included within the main sublineage of lineage B, with little phylogenetic structure evident. This is not what would be expected under the hypothesis of palaeodrainage connectivity, in which it would be expected that close relationships would be seen between populations previously connected and greater divergence between populations that have never been connected via these historic drainages.
Only one Haplotype (Hap 4) was found within lineage A (Fig. 5), expanding the distribution of this lineage observed by McCluskey to include Deep Creek in the Maribyrnong drainage basin which is located at the top of Port Phillip Bay (Fig. 6A). The range observed by McCluskey included Bunyip, East Gippsland, South Gippsland, Snowy, Mitchell, Tambo, Loddon and Inman SA. No individuals within this study were found within McCluskey’s lineage C (Fig. 5) which was found mainly across inland Victoria and south east South Australia (Fig. 6C) or lineage D (Fig. 5) which was found in the coastal Glenelg drainage and inland Wimmera drainage in far west Victoria (Fig. 6D). Five haplotypes (Haps 1, 2, 3, 5 and 6) (Fig. 5) were divergent from all other samples observed in this study and do not belong to any of the clades found by McCluskey (2007) (for location of these lineage refer to Fig. 6D). The divergent haplotype 1 (Yarra) forms a sister group with lineage D (Glenelg and Wimmera), however they were not seen to be closely related due to high levels of genetic diversity (6%; Table 2) between haplotype 1 and lineage D, double that which is found within the lineage. The remaining haplotypes are clearly divergent from each other (Fig. 5, Table 2) and the named lineages. These 20
represent new lineages within Victoria although, due to low support values on the branches, their phylogenetic affinities are unclear.
Table 2 Kimura 2 parameter (K2P) genetic distances among lineages found in this study. Values along the horizontal (in bold) represent the average genetic divergences within each lineage. 1 2 3 4 5 6 7 8 9 10 1 P. howensis Lineage A 0.19 Lineage B 0.20 Lineage C 0.19 Lineage D 0.21 Hap 1 0.19 Hap 2 0.20 Hap 3 0.20 Hap 5 0.20 Hap 6 0.20 2 0.01 0.08 0.07 0.07 0.07 0.08 0.10 0.08 0.08 3 4 5 6 7 8 9 10
0.02 0.10 0.09 0.08 0.08 0.09 0.08 0.08
0.00 0.09 0.09 0.11 0.10 0.08 0.08
0.03 0.06 0.08 0.08 0.08 0.08
0.06 0.07 0.07 0.07
0.08 0.08 0.08 0.08 0.08 0.00 -
21
Lineage C
Lineage A
Lineage B
Lineage D
Figure 5 Bayesian phylogenetic analysis of P. australiensis. Bayesian posterior probability values above 0.95 shown above the branches, Maximum Likelihood bootstrap values above 70 shown below the branches. For the distribution of lineages refer to Fig. 6 over page. The colours (yellow, purple, pink and green) correlate to colours in figure 6D. Haplotypes found by this study are identified with an *.
22
Figure 6 Geographic distribution of each of the P. australiensis lineages observed in the phylogenetic analysis. Maps A, B and C represent these lineages respectively. Map D represents both lineage D (black) and four divergent lineages. Colours (green, pink, blue and purple) correspond to those found in figure 5.
23
3.1.3 Haplotype Network Figure 7 shows the geographic distribution of the 21 haplotypes in lineage B observed in this study. Very little geographic structure is observed, with many haplotypes being present across diverse locations. The ancestral haplotype (Hap 27) is the most broadly distributed haplotype observed in this study, being found across Hopkins, Otway Coast and South Gippsland drainage basins. As is clear in Figure 4, no rivers between these drainage basins have ever been connected via palaeodrainage networks. Hopkins and Otway Coast drainage basins are adjacent and have a distance of about 33km between the two closest sites sampled (Hopkins River and Curdies River) but South Gippsland is geographically distant being separated by both Port Phillip Bay and Western Port Bay.
The tip clades shown in the network represent more recent divergence which is spread across almost all drainage basins sampled, Barwon, Werribee, Maribyrnong (from the Port Phillip Drainage) and Coal River Tasmania drainages are all represented in more recently diverged tip clades, but not in the main ancestral clade, suggesting more recent population expansion in these more central locations. Haplotype 9 is the most divergent haplotype from this study that fell within lineage B. It is found in the more central drainage basins, including Werribee, Maribyrnong, Yarra and Barwon that were not represented in haplotype 27.
24
Figure 7 Median Joining Haplotype network of P. australiensis lineage B. Pie charts show drainage basins where each haplotype is located. 25
3.1.4 Summary Statistics Maribyrnong, Barwon and Werribee all had much higher nucleotide (π) diversity than any other population (Table 3), which shows that there is greater genetic diversity found around the more central coastal area, which is due to the appearance of the unique sub lineage within haplotype 9. These three populations are also the three populations found in tip clades of the haplotype network (Fig. 7) that were not represented in the ancestral clade. No haplotype (h) or nucleotide diversity was found in samples collected from the Yarra drainage (Table 3). This may mean that there is actually little diversity within the Yarra, however more likely as the samples were collected from one location the population collected, consisted entirely of siblings and is not a true representation of the diversity within this site. The overall haplotype and nucleotide diversity is not greatly different from that of individual populations, showing there are no great differences across the species range.
Table 3 Population level summary statistics for lineage B of P. australiensis. Shown are number (n) of individuals, number (n) of haplotypes, number of segregating sites (S), haplotype diversity (h) and nucleotide diversity (π) Population All Yarra Maribyrnong Barwon Werribee South Gippsland Otway Coast Hopkins Coal River n (individuals) 83 9 4 5 6 8 26 21 4 n (haps) 21 1 2 2 4 3 6 7 2 S 31 0 12 15 16 2 4 10 1 h 0.839 0.000 0.667 0.400 0.867 0.464 0.612 0.786 0.500 π 0.008 0.000 0.014 0.013 0.014 0.001 0.001 0.003 0.001
26
Within the demographic stability summary statistics only three significant results were observed (Table 4). These included a negative Fu and Li’s F* value found for the Barwon drainage basin. A negative Fu’s Fs value was found for Otway Coast drainage and a positive Fu’s Fs found for Coal River. Due to the lack of genetic diversity found within the Yarra drainage basin it could not be included in the demographic stability summary statistics (Table 4). These results are indicative of population growth in Barwon and Otway Coast drainages, and a population bottleneck in Coal River. They are however based only on one significant statistic per population out of the five measures used, which would indicate a lack of support for these significant results. Overall the lack of significant results suggests that the populations all remained demographically stable.
Table 4 Demographic stability summary statistics for drainage basin populations. Tajima’s D (Tajima 1989), Fu and Li’s F* and D* (Fu and Li 1993), Fu’s Fs (Fu 1997) and the R2 statistic (Ramos-Onsins and Rozas 2002). * p
