Results
Assay efficiency, sensitivity, and
specificity
Amplification efficiency was 99%
and 96% with an R 2 value of 1.0 for theL. sativae and L. brassicae real-time qPCR assays,
respectively. The LOD was 0.0056 pg (0.0035 pg) of DNA and the LOQ was
0.022 pg (0.017 pg) of DNA for the L. sativae (L.
brassicae ) assay.
There was no off-target amplification of the seven Liriomyzaspecies for either assay, highlighting the specificity of these
real-time qPCR assays.
Experiment 1 – Testing of unmined leaves in the field
During the 2018 trials, L. sativae was diagnosed in 1 out of 10
of the unmined leaves, with a low DNA concentration found (1 pg DNA in 2
of 3 replicate qPCR tests). DNA concentrations from the ‘positive
control’ samples averaged 123 pg across 71 samples tested (Figure 1).
During the 2019 trials, L. sativae was diagnosed in 2 out of 45
unmined leaves, with low DNA concentrations detected (<2 pg
DNA in all three replicate qPCR tests). The two leaves with a low (but
positive) diagnosis were collected from areas with known L.
sativae activity, FGG and GHF. However, DNA of L. sativae was
not amplified in any of the INJ unmined leaves, where L. sativaehas never before been recorded.
Experiment 2 - eDNA persistence
trial
Temperatures recorded over the
period during which the leaf mines were ageing ranged from 20.1 °C to
37.6 °C (averaging 29.5 °C in the daytime and 26.7 °C overnight) and
relative humidity levels ranged from 33.8% to 98.7% (averaging 64.3%
in the daytime and 73.6% overnight). While the amount of DNA amplified
varied widely with age of leaf mines (Figure 1), DNA was still
diagnosable within leaf mines after 28 days, and the average success
rate for diagnosis from an empty leaf mine (a successful diagnosis being
the case where at least one of three replicate qPCR tests for an
individual leaf mine yielded positive results) for mines between 0 and
28 days old was 65.1% (Figure 2). There was no effect of age of the
leaf mine (F = 2.5, df = 1, p = 0.12) on the concentration
of DNA amplified from the leaf material.
Experiment 3 – eDNA sensitivity under field conditions
Of the 184 leaf mine samples from M. atropurpureum directly
preserved in ethanol, 82.1%
yielded a positive diagnosis of L. sativae (on at least one of
three replicate qPCR tests) and 17.9% yielded no diagnosis. Of the 104
leaf mines preserved onto FTA cards, 58.7% yielded a positive diagnosis
of L. sativae (on at least one of three replicate qPCR tests) and
41.3% yielded no diagnosis.
There was no effect of leaf mine age (F = 0.5, df = 2, p = 0.60)
or leaf mine length (F = 2.5, df = 2, p = 0.09) on the
concentration of DNA amplified from the leaf material. However, there
was a significant effect of larval remains on the concentration of DNA
amplified within a leaf mine (F = 3.5, df = 8, p< 0.001). For mines preserved in ethanol, those containing the
remains of a fly larva amplified on average 488.5 (+ 116) pg of
DNA compared with 36.4 (+ 22) pg from mines that did not contain
larval remains. For mines preserved onto FTA cards, those containing
larval remains amplified on average 10.5 (+ 10) pg of DNA compared
with 0.09 (+ 0.03) pg for those mines that did not contain larval
remains. For leaf mines preserved in ethanol that did not contain larval
remains, 89.4 % yielded a positive diagnosis.
Using the multilevel detection probability model incorporating
uncertainty at the eDNA and qPCR stages, it was estimated that the
probability of L. sativae diagnosis increased with the number of
mined leaves and technical replicates tested as shown by the positive
coefficients in Table 2 (with the exception of FTA storage). If DNA was
present in a leaf mine, each qPCR replicate had an estimated 80% chance
of confirming presence (with three technical replicates providing over
99% chance of confirmation). Of the covariates tested, the presence of
larval remains and the preservation method were found to have
significant effects on the probability of L. sativae diagnosis
(Table 2). The presence of larval remains increased the probability of
diagnosis by 9 times, while the use of FTA cards for preservation
decreased the probability by 5 times compared with preservation of empty
leaf mines in ethanol (Table 2).
Assuming three technical qPCR replicates and samples are preserved in
ethanol, testing of only two empty leaf mines is estimated to provide a
90% probability of L. sativae diagnosis, even if there are no
larval remains present, the leaf itself is not fresh, and the mine is 50
mm (or less) in length (Figure 3). Three empty leaf mines provided a
greater than 90% probability (as indicated by the 95% credible
interval in Figure 3)
Experiment 4 - Field applications to delimit geographic
range and host range
The empty mine eDNA method was applied to several plant samples
collected within, and outside of the known range of L. sativae in
Australia. Of the mines collected from known host plants of L.
sativae on Thursday Island, all collected mines from S.
lycopersicum and a single mine from V. unguiculata were
confirmed to be L. sativae (Table 3). Of the mines collected from
plants which were not known to be a host L. sativae in the Torres
Strait, namely Capsicum sp., P. edulis , T.
ulmifolia , O. basilicum and S. jamaicensis , at least one
mine per host plant was confirmed to be L. sativae . This
constitutes the first formal record of mining activity of L.
sativae in these plants in Australia.
Of the mines collected outside the known range of L. sativae ,
five (of the seven) leaf mines from M. atropurpureum collected on
Zuna Island were confirmed as L. sativae , while the leaf mine
collected from S. melongena in Cairns was not found to beL. sativae (Table 3)