Discussion
The eDNA method developed here was shown to be highly robust under field
conditions, reliably detecting and diagnosing L. sativae DNA from
empty leaf mines under tropical conditions. This novel approach
significantly increases the surveillance opportunity for this invasive
pest and is likely to be highly transferable to other globally important
agromyzid flies. Preservation methods of collected leaf samples were
found to be important with 100% ethanol being the most reliable
collection procedure (but FTA cards were also feasible). There was no
decline in diagnostic success measured up to 28 days after the emergence
of larvae from a mine. Other forms of arthropod eDNA, such as that of
the tiger mosquito (Aedes albopictus) , have been shown to persist
in the environment for similar lengths of time (Schneider et al. 2016).
Preliminary field applications have already highlighted the utility of
the eDNA method through identification of new local hosts, includingS. jamaicensis , C. annuum , T. ulmifolia, P. edulisand O. basilicum , and the first occurrence record of L.
sativae on Zuna Island. These results widen the range of hosts that
will be considered during routine surveillance aimed at containing this
pest. While the current preference for these host plants appears low,
the occurrence patterns of these plants could nevertheless make them
important vectors of spread. Passiflora edulis , C. annuumand O. basilicum are common garden plants, with O.
basilicum posing a particular risk given it is a popular herb that is
routinely transported by tourists and locals as living plants.Stachytarpheta jamaicensis is a very common and widely
distributed weed, particularly in disturbed areas throughout northern
Queensland and other parts of Australia (Atlas of Living Australia
2019). This new knowledge has already been used by the Northern
Australian Quarantine Strategy in their regular surveillance programs
for L. sativae (B. Waterhouse and S. Cowan, pers. comm.).
Out of 45 unmined leaves, two ‘false positives’ occurred (with
<2 pg DNA detected in all three replicate qPCR tests). It is
unclear whether these were false positives in the true sense (as a
result of off-target DNA being amplified by the eDNA assay), whether
they were the result of laboratory contamination (despite careful sample
hygiene where negative control leaves were collected and processed in a
different laboratory than mined leaves and with heat sterilized
equipment), or whether there was indeed L. sativae DNA already
present on the unmined leaves. Both ‘false positives’ came from unmined
leaves that were collected from sites with adult L. sativaeactivity present, FGG and GHF. The lack of any false positives from INJ,
the only site included with no known presence of L. sativae ,
increases our confidence in the reliability of the eDNA assay. False
positives can be a concern within a biosecurity surveillance program.
For instance, a false positive could trigger unnecessary economic losses
as a result of unwarranted trade restrictions, while a false negative
could delay the detection of an incursion of a pest or disease, reducing
the odds of eradication and increasing the costs of control (Fleming et
al. 2017; Epanchin-Niell and Hastings 2010; Moore et al. 2010). Thus,
the cost of errors must be weighed alongside the goals of the
surveillance program and the resources available (Garrard et al. 2008).
For early detection surveillance, a less conservative survey method or
diagnostic test that aims to reduce the chances of false negatives may
be preferred. Similarly, equivocal eDNA assay results (where one or two
technical replicates out of three yield a positive detection) should be
considered in a similar manner, particularly in areas where the target
species has not been detected previously. Given biosecurity response
decisions are typically based on a high level of confidence, equivocal
results would serve as an indicator that further surveillance and
diagnostic testing is warranted.
In this study, we developed and tested the in-field reliability and
practicality of an eDNA assay for leaf mines (Experiments 1 through 3),
and applied this to better understand the dynamics of the current
incursion of L. sativae within Australia (Experiment 4). Leaf
mined samples tended to be quite rare in our field surveys and thus
represented a relatively small sample size, sometimes being comprised of
only a single leaf mine (Table 3). This is to be expected in early
incursions undergoing range and host expansions (Liebhold et al. 2020).
Nonetheless, analysis of samples collected during our survey indicate
expansions in the known geographic and host range of L. sativae .
The eDNA assay therefore has the potential for detecting incursions at
an early stage, which is more likely to increase the success of a
biosecurity response (Fleming et al. 2017; Epanchin-Niell and Hastings
2010; Moore et al. 2010). The identification of a new L. sativaehost plant within the Torres Strait, S. jamaicensis , is
particularly noteworthy and has already directed surveillance programs
being undertaken in Australia by government agencies.
The eDNA method developed herein is likely to become an important tool
for distinguishing agromyzids that remain a biosecurity threat to
Australia’s agricultural
industries. It presents a new
opportunity to reduce the spread of exotic leafminers, which,
historically, have been difficult to detect and contain overseas due to
poor host records and overlap with native leafminers (Powell 1981). For
example, not only is the leaf mining damage created by L. sativaeindistinguishable from other high-risk exotic agromyzids, includingL. trifolii and L. huidobrensis , it can also be easily
confused with the closely related L. brassicae , already common in
Australia, which not only creates indistinguishable damage on many of
the same plants as L. sativae , but is also indistinguishable as
an adult by any means other than dissection or molecular analysis (Shiao
2004; International Plant Protection Convention 2016).
The utility of eDNA in biosecurity programs will extend beyond
leafmining species. For example, eDNA approaches using samples collected
from water bodies have greater sensitivity compared with traditional
surveillance techniques for invasive mosquito larvae (Aedes
albopictus , A. japonicus japonicus and A.
koreicus ) (Schneider et al. 2016) and the invasive American bullfrog
(Rana catesbeiana ) (Dejean et al. 2012). Yet, there remain many
unexplored opportunities to apply eDNA to enhance detection of exotic
species, particularly in the terrestrial realm. Recently, eDNA
approaches have been developed to detect the highly invasive brown
marmorated stinkbug (Halyomorpha halys ) in orchards from water
reservoirs used to wash fruit as well as bat faecal material; both
methods achieved a higher sensitivity than traditional trapping methods
such as light and pheromone traps (Maslo et al. 2017; Valentin et al.
2018). Opportunities for early detection of other exotic invertebrate
pests also exist and remain to be explored. For instance, the giant
African snail (Lissachatina fulica ) leaves a trail of slime that
is likely to include their DNA and developing an efficient eDNA sampling
protocol could be a more sensitive method of detection at borders (e.g.
sea ports). Beyond early detection of exotic species, scalable and
high-throughput eDNA techniques should become increasingly valuable for
routine monitoring programs. Such technologies will increase the
efficiency and sensitivity of both delimiting responses and proof of
area freedom assessments, where the large-scale sampling necessary for
statistical reliability is often associated with large time and labour
costs (Gambley et al. 2009; Abdalla et al. 2012).
Although promising, eDNA will not always result in reliable diagnostics.
Derocles et al. (2015) used a mini-barcoding approach to target degraded
DNA inside empty leaf mines in a number of plant hosts, but found only
6% of the mines collected from the field (and therefore of an unknown
age) allowed for species identification, and only 33% of the mines with
a known age (where the emergence of the leafminer was observed and
recorded) allowed for identification. This is far lower than the
detectability we found for L. sativae . In part, this may be
attributed to the use of general primers (designed for a range of
lepidopteran and dipteran species) and PCR conditions not optimised to
highly degraded DNA (Derocles et al. 2015). Within our study, a highly
specific qPCR assay for L. sativae was applied that could detectL. sativae DNA quantities as low as 0.1 pg. The use of specific
primers, qPCR detection, as well as improved DNA extraction methods is
likely to have accounted for the higher detection rates compared with
previous studies.
In many cases, eDNA approaches may not replace traditional surveillance
techniques, but rather complement them for improved detection (Schneider
et al. 2016). In situations where false positives could lead to
considerable economic losses (e.g. trade restrictions being imposed),
morphological support should ideally complement molecular tests.
Further, the success of an eDNA approach will be dependent on the target
species lifecycle and biology. To determine the potential for a
successful eDNA approach, one must consider what traces a target species
might leave behind in its environment, and where those traces are most
likely to be concentrated. For example, species with highly localised
activity, such as pollinating insects (Thomsen and Sigsgaard 2019), may
be more conducive to such an approach than transient species. However,
it may be possible to improve detection rates for transient species via
‘eDNA traps’ that take advantage of the target species’ behaviour or
industry practices that concentrate eDNA. Burns et al. (2020) employed a
novel eDNA trap to detect the threatened Baw Baw frog (Philoria
frosti ) and its’ key threatening pathogen chytridiomycosis. The
application of eDNA in terrestrial environments is challenging, but
innovative approaches can result in achieving methods being developed
that are more sensitive and efficient than traditional approaches.
As the interconnectedness of global economies continues to remove
spatial barriers between regions, biosecurity technologies must keep
pace to mitigate the enormous pressures from alien species placed on
biological systems. Novel eDNA methods based on modern technologies can
play an important role in modern biosecurity efforts to protect both
natural and agricultural ecosystems.