Screening mitochondrial DNA sequence variation as an alternative method for tracking established and outbreak populations of Queensland fruit fly at the species southern range limit

Abstract Understanding the relationship between incursions of insect pests and established populations is critical to implementing effective control. Studies of genetic variation can provide powerful tools to examine potential invasion pathways and longevity of individual pest outbreaks. The major fruit fly pest in eastern Australia, Queensland fruit fly Bactrocera tryoni (Froggatt), has been subject to significant long‐term quarantine and population reduction control measures in the major horticulture production areas of southeastern Australia, at the species southern range limit. Previous studies have employed microsatellite markers to estimate gene flow between populations across this region. In this study, we used an independent genetic marker, mitochondrial DNA (mtDNA) sequences, to screen genetic variation in established and adjacent outbreak populations in southeastern Australia. During the study period, favorable environmental conditions resulted in multiple outbreaks, which appeared genetically distinctive and relatively geographically localized, implying minimal dispersal between simultaneous outbreaks. Populations in established regions were found to occur over much larger areas. Screening mtDNA (female) lineages proved to be an effective alternative genetic tool to assist in understanding fruit fly population dynamics and provide another possible molecular method that could now be employed for better understanding of the ecology and evolution of this and other pest species.


| Bactrocera tryoni distribution and management
Bactrocera tryoni occurs along the entire east coast of Australia from the tropics in Queensland to temperate eastern Victoria (Dominiak & Daniels, 2012). Bactrocera tryoni is restricted to Australia (endemic) and some Pacific islands and is of significant trade concern for national and international export of some horticultural products (Plant Health Australia 2016). The history of B. tryoni control was recently reviewed by Dominiak and Ekman (2013). Large production areas in southern Australia benefit from market access opportunities and avoided production losses through being free of this significant pest (Clarke et al., 2011). The main horticultural production areas across Victoria, New South Wales, and South Australia in southeastern Australia-were until very recently covered by a large Fruit Fly Exclusion Zone (FFEZ)-developed in 1995 ( Figure 1), with procedures to manage the FFEZ and permitted control measures (Dominiak & Daniels, 2012). A second zone of higher quarantine management, the Greater Sunraysia Pest Free Area (GSPFA), was established in 2007 within the FFEZ ( Figure 1). The aim of the GSPFA was to optimize international market access for local stone fruit, table grapes, and citrus producers. The GSPFA consists of a long zone following Australia's largest river, the Murray River, while the FFEZ involved a larger area encompassing large areas of semiarid landscape, the latter largely not supporting the survival of B. tryoni (Dominiak, Mavi, & Nicol, 2006). A further informal Risk Reduction Zone (RRZ) also existed at the boundary of the FFEZ and established region (Figure 1). Management within these control zones is based on surveillance (Dominiak, Gilmour, Kerruish, & Whitehead, 2003;Dominiak & Nicol, 2010). When detection exceeds a prescribed threshold , an outbreak is declared and control measures immediately instituted, including spraying and restrictions on the movement of locally grown host produce (Dominiak & Ekman, 2013). However, recently, the pesticides used in cover sprays in some eradication programs have been reviewed and their use patterns are now much more restricted (Dominiak & Ekman, 2013).

| Molecular markers for B. tryoni surveillance
To achieve the most effective control of the spread of B. tryoni, a major concern is tracking fly dispersal, including understanding the origins of new incursions/outbreaks. A large number of nuclear genetic markers F I G U R E 1 Geographic distribution of Bactrocera tryoni at the species range limit in southeastern Australia (adapted from Dominiak & Daniels, 2012), with the boundaries of the fruit fly management control regions mentioned in the text indicated. The red box indicates the area sampled in this study (Victoria and southern New South Wales) (microsatellites) have been developed for B. tryoni (Kinnear, Bariana, Sved, & Frommer, 1998;Wang, Yu, Raphael, & Gilchrist, 2003;Zhou, Frommer, Sved, & Gillies, 2003), which have been employed to examine underlying population structure in this species (Cameron, Sved, & Gilchrist, 2010;Gilchrist, Dominiak, Gillespie, & Sved, 2006;Gilchrist & Meats, 2010;Wang et al., 2003;Yu et al., 2001;Chen, Dominiak, & O'Rourke, 2016), including the structure and persistence of B. tryoni outbreak populations. Many additional nuclear markers, including thousands of SNPs, were recently characterized from the B. tryoni genome (Gilchrist et al., 2014); however, these are yet to be employed to characterize B. tryoni populations (Sherwin et al., 2015).
Previous microsatellite studies have indicated the presence of a number of genetic populations in southeastern Australia, including a genetic cline between two populations in the RRZ along the border of the established and managed zones (Gilchrist & Meats, 2010).
Recently, the use of microsatellites for management of B. tryoni outbreaks has been significantly improved through the optimization of nine loci to work together in a single multiplex PCR (Chen et al., 2016), greatly reducing the labor required to genetically screen specimens. However, the use of microsatellite data for genetic screening does still have some limitations including the following: (1) They can require relatively large sample sizes to define populations (e.g., Gilchrist, Sved, & Meats, 2004); (2) they are also not particularly suited to the cumulative addition of samples, owing to potential differences in allele scoring between different laboratories and genotyping platforms (e.g., Moran, Teel, LaHood, Drake, & Kalinowski, 2006).
An alternative molecular marker, mitochondrial DNA (mtDNA), that is inherited independently from nuclear (microsatellite) markers has previously been employed for species identification of B. tryoni (Armstrong & Ball, 2005;Blacket, Semeraro, & Malipatil, 2012;Cameron et al., 2010). MtDNA DNA sequences are ideal for cumulative studies, as they can be unambiguously scored and can be built up into a spatial and temporal "library" as samples become available.
MtDNA sequences are also relatively cheap to obtain and simple to analyze, in comparison with nuclear markers such as microsatellites.
To date, a variety of different mtDNA loci have been examined (Blacket et al., 2012;Cameron et al., 2010;Morrow et al., 2000;Shearman, Frommer, Morrow, Raphael, & Gilchrist, 2010). However, none of these studies have included a large number of samples of B. tryoni from southern Australia at the species natural range limit (Dominiak & Daniels, 2012). Some fruit fly species are known to be limited by environmental conditions (e.g., Royer, Wright, & Hancock, 2016), and southern Australia is believed to be suboptimal for B. tryoni (Gilchrist & Meats, 2010;O'Loughlin, 1984), with temperature, availability of moisture and host fruit believed to be the major factors determining the suitability of areas for B. tryoni (Clarke et al., 2011;Dominiak et al., 2006). Previous studies suggest that some of the most southerly known established populations, from east Gippsland Victoria (Figure 1), may have adapted to colder environmental conditions (O'Loughlin, 1964).

| Objectives
The primary aim of this study was to assess the utility of using DNA sequences from a single mtDNA marker (i.e., haplotypes) to detect the underlying population structure within and between established and outbreak B. tryoni populations within different fruit fly management control regions ( Figure 1) in southeastern Australia (i.e., spatial variation). We also examined haplotype variation over time (i.e., temporal variation) at a small number of selected sites that were historically subject to different B. tryoni control measures.

| Adults
A total of 765 adult specimens from 63 locations were examined in this study ( Table 1). All of these specimens were adult male B. tryoni  Table 1). The above a priori management control regions were used in the analyses below as a convenient way of grouping samples that differ in the degree of effort applied to control B. tryoni populations (i.e., "Management zone" in Table 1).

| Larvae
Additionally, a small number of larval samples were examined in this study. One group of larval samples originated from Queensland (n = 12) that were collected by Victorian (Agriculture Victoria) Biosecurity from fruit intercepted at the Melbourne markets. The other larval samples (n = 13) were collected from infested fruit grown in the GSPFA during outbreaks, from sites that were also sampled contemporaneously for adult B. tryoni (Table 1). Each larval sample tested represented a different larval detection (i.e., a separate infested fruit sample).

| Spatial analyses
Spatial autocorrelation was used to detect relationships between genetic and geographic distances within regions, in GenAlEx (Peakall & Smouse, 2006). Geographic distance matrices (km between sites) were estimated in GenAlEx from the latitude and longitude of each site. Genetic distance matrices were calculated for a haploid marker by population in GenAlEx. To confirm the robustness of any significant correlations, multiple distance classes (between 5 and 100 km) were trialed for each spatial autocorrelation analysis (data not shown), and no other statistical corrections were made to account for the large number of spatial autocorrelation comparisons tested.

| Genetic analyses
The genetic diversity present at sites and within regions was examined by plotting the number of haplotypes detected at sites, divided by the number of individuals sampled at each site, against the latitude and longitude of each site, in Excel. An analysis of molecular variance (AMOVA) was also conducted in GenAlEx to examine the partitioning of genetic variation within and between regions and sites.
A neighbor-joining tree of genetic relationships between sites was constructed in Mega 5.  conducted here as these would be inappropriate for these data given that most sequence differences between haplotypes are likely to have accumulated over long periods of time in the original northern source populations, (which have not been sampled for mtDNA variation to date see below), rather than in situ in southeastern Australia.

| Temporal analyses
Finally, the genetic (haplotype) diversity in samples collected from a limited number of sites in Gippsland and the FFEZ in 2008 and 2010 was examined to test for the persistence of mtDNA lineages through time, that is, temporal variation at these sites.

| Overall genetic diversity
A large number of haplotypes (n = 153) were detected from the B. tryoni samples in this study (Table 1). The AMOVA indicated that the greatest amount of genetic variation (74%) was present within populations (sites), a substantial amount of variation (24%) was found between populations (sites), while a very small amount (2%) was limited to regional differences. Despite the very low regional distinction indicated above, further analyses were conducted using the a priori management regions defined earlier, as these differ not only in the degree of control measures (e.g., low in established regions compared to high in outbreak areas), but also in their observed levels of observed genetic diversity and levels of gene flow (see results below).

| Genetic diversity within sites
A comparison of haplotype diversity (i.e., number of alleles/number of samples at a site) compared with site locations was conducted ( Figure 2). A strong association (regression analysis) was detected with longitude but not latitude, with each management region also exhibiting different levels of diversity ( Figure 2). Sites in the GSPFA possessed low haplotype diversity, FFEZ sites were variable with low to high diversity, while established sites all exhibited high genetic diversity.

| Spatial autocorrelation within regions
Spatial autocorrelation analyses were employed to test for possible

| Genetic relationships between sites
The analysis of genetic relationships (based on the degree of haplotype sharing) between sites sampled in this study (Figure 4) The geographic extent of each of these major genetic groups is illustrated in Figure 5a, and the most common haplotypes detected in each population are listed in Table 2

| Temporal patterns of variation
Other indirect evidence for adult B. tryoni breeding at outbreak sites is provided by the temporal samples examined from Gippsland  Table 1). The presence of the same haplotypes between years at outbreak sites might indicate persistent recolonization from the same source population, or could be due to persistent residual populations at these sites. The latter possibility appears more likely when outbreaks are present continuously, that is, over a number of consecutive years, given that B. tryoni are known to have more than one generation a year (Clarke et al., 2011;O'Loughlin, 1984). The lack of shared haplotypes at the other two Gippsland sites (Eagle Point and Sale) appears to indicate nonpersistence of populations and probably different outbreak sources between the years sampled.

| Comparison of mtDNA haplotype variation with previous molecular data
Compared with previous multilocus nuclear microsatellite marker studies of B. tryoni populations (Chen et al., 2016;Gilchrist & Meats, 2010;Gilchrist et al., 2004Gilchrist et al., , 2006, mtDNA provides information from only a single locus, which has a clonal-maternal mode of inheritance. Differences between these two sets of markers could result in divergent patterns of population genetic variation being detected. For example, all of the samples in the current study were males, so we were unable to examine whether there were sex-specific differences in dispersal, whereas microsatellites would provide genetic information from F I G U R E 4 Neighbor-joining tree of genetic relationships between sites, based on the degree of haplotype sharing (Nei distances), between all Bactrocera tryoni collection sites (locality codes from Table  1). Major genetic groups are indicated by colored boxes (dashed lines) and are named after the geographic region in which they generally occur ( Figure 5). Greater Sunraysia Pest Free Area sites are highlighted by an "*", filled circles next to each population indicate a priori management regions from Figure 1 Berr   Table 1) and simple analyses, we found highly congruent results between the mtDNA data presented here and the previous results from nuclear microsatellite markers, with very similar resolution of underlying B. tryoni population genetic structure observed between studies.
Similarities with previous studies include the following: (1) (Figure 2; Gilchrist & Meats, 2010;Gilchrist et al., 2004Gilchrist et al., , 2006; (5) many of the southeastern Australian sites sampled in both sets of studies appear genetically distinct; for example, the Sydney region was found to be dissimilar to inland NSW for mtDNA (our study) and microsatellites (Gilchrist et al., 2006).

| Patterns of mtDNA variation
Overall the spatial analyses of southern B. tryoni populations suggest a degree of localization of genetic variation within a number of outbreaks (Table 1, Figure 5a). Examinations of the most common mtDNA haplotypes within each area show that there is a different "suite" of common haplotypes within each, with a single dominant haplotype accounting for almost a third of all haplotypes in these populations ( Table 2). The Geographic groups have been selected based on genetic similarities indicated in Figure 4. Only haplotypes that account for more than 5% individuals within each group are listed here. FFEZ, Fruit Fly Exclusion Zone.

F I G U R E 5 (a) Collection localities of
Bactrocera tryoni (locality codes from Table  1), indicating regions and management zones, approximate boundaries between zones (dashed lines), and approximate location of long-term roadblocks (black rectangles). Major genetic groups (from Figure 4)  Movement of B. tryoni between sites is described as jump diffusion, that is, the human-assisted long-distance movements followed by local diffusion through natural insect flight (Sadler, Florec, White, & Dominiak, 2011). The natural dispersal ability of B. tryoni is generally less than one kilometer (Dominiak, 2012), while B. tryoni larvae are known to be carried in infested fruit (Dominiak & Coombes, 2009;Dominiak & Daniels, 2012;Dominiak, Rafferty, & Barchia, 1998 (Figures 4 and 5). Wagga Wagga is the second largest inland town in NSW but is more isolated, not being on a major throughway, although it is on a road system of east-west traffic from Sydney (B. tryoni established) into the eastern FFEZ and subsequently to the GSPFA and South Australia (the latter three areas usually being B. tryoni free).
Interestingly, Wagga Wagga appears more similar to some FFEZ sites, more closely matching sites from the "Central FFEZ" (Figures 4 and 5) than to other nearby towns in NSW. Albury/Wodonga is also a large town situated on the major north-south transport route taking traffic from Sydney (B. tryoni established) to Melbourne (B. tryoni free), which appear to match the adjacent "Southern FFEZ" sites ( Figures 4 and 5), even though road traffic largely does not enter the FFEZ.
Ours is the first genetic study to examine the most southerly B. tryoni populations, including samples obtained from Gippsland ( Figure 1, Table 1).
These southern established populations, whose existence has been known of for at least fifty years, may be adapted to local conditions (O'Loughlin, 1964). In our analyses, Gippsland flies did not appear distinctive overall from other established Victorian or NSW sites (Figures 4 and 5a). However, interestingly Gippsland sites did possess a large number of haplotypes (n = 15,Haps 42,45,48,58,65,75,79,88,107,129,130,150,157,158,159) that were not found elsewhere in our study. It would be useful in future studies to sample a greater range of B tryoni genetic diversity from a greater part of the species range, including sites from the northern parts of the species range, for comparison with the data presented here.

| Persistence of populations
At the time of our study, outbreaks in some parts of the southeast-  2, 3, and 5). Most outbreaks in other parts of the FFEZ and GSPFA were only declared during the study period, and these sites appear to generally exhibit variable (rather than continuous) gene flow.
The persistence of some populations within parts of the FFEZ was also supported by our limited examination of temporal variation ( Figure 6).
If there was a widespread low-level resident population within areas of the GSPFA prior to the study period, gene flow between sites would be expected to result in some shared common haplotypes across the GSPFA. This does not appear to be the case, and overall genetic patterns appear to indicate that over the study period, there were many independent introductions of B. tryoni into the GSPFA from multiple sources. Furthermore, these introductions appeared to show very limited dispersal between GSPFA sites. Florec, Sler, White, and Dominiak (2013) (Dominiak et al., 1998), some of which is infested, with these fruit movements creating the long-distance jump dispersal. Historically, there has been an active random vehicle inspection (see Figure 5a) and a community awareness program at the eastern border of the FFEZ in an attempt to minimize these jump dispersals (Dominiak & Coombes, 2009. Our results indicate that these activities were not entirely successful, especially within the RRZ. A similar but smaller program was also run on the eastern border of the GSPFA (see Figure 5a). Our results do indicate a degree of effective control of B. tryoni movement between many geographically close sites, with numerous adjacent sites in the FFEZ and GSPFA being genetically dissimilar, that is, not from the same outbreak source.

| Pest management implications
Historically, the management of B. tryoni in the FFEZ and GSPFA has relied on insecticide cover sprays; however, the use of some pesticides has been recently restricted (Dominiak & Ekman, 2013). Fruit fly management has become more challenging, with growers now required to follow an area-wide management approach (Florec et al., 2013;Lloyd et al., 2010). Following the wettest two-year period on record in the FFEZ (Webb, 2012), eradication of more than one hun- However, any reduction in the current vehicle inspection program on vehicular traffic entering South Australia will increase the risk of incursion (Florec et al., 2013).
Our study provides a snapshot of a particular point in time, when B. tryoni was in the process of becoming established at the southernmost extremity of the species geographic range. This study illustrates how the molecular monitoring methods employed here, to better understand pest dispersal, could be applied to other pests in production areas that are trying to develop pest-free places of production, or areas of low pest prevalence (Dominiak et al., 2015), or in emergency management of recently incurring pests (e.g., Blacket et al., 2015).

| CONCLUSIONS
Overall, it appears that screening mtDNA (female) lineages does provide an extremely useful alternative tool for helping to understand B. tryoni population dynamics and determining possible sources of outbreaks. In this study, the extent of gene flow was estimated across regions and a number of distinctive populations were detected within the FFEZ and GSPFA, with some genetic lineages appearing to be being maintained (i.e., breeding), while others appeared to have not yet become widespread and persistent at the time of the study.
We provide baseline information on a number of new introductions of B. tryoni at the southern extremity of the species range, that could now be built upon, screening additional sites to include a greater part of the range of B. tryoni (i.e., sampling a larger part of the overall B. tryoni genetic diversity) as well as examining the same sites over time. It would be particularly beneficial to revisit the sites sampled in our study to observe the genetic changes that have occurred since control measures were reduced in the FFEZ and GSPFA.
Future DNA sequencing studies will now be relatively easier to implement using newly available next-generation high-throughput  This work on southern Bactrocera tryoni was supported through a Victorian Department of Primary Industries (Agriculture Victoria) funded pilot study (Project 103890) and through funding from NSW DPI. Agriculture Victoria and NSW DPI staff read and approved this manuscript and provided useful comments and suggestions.