1. Incidence of parasitism is often underestimated because ‘ghost’ parasitoids (deadnunemerged parasitoids or those that have emerged leaving the host carcasses) are difficult to detect and identify. This study demonstrates that the use of melting curve analysis (MCA) of host carcasses can detect and identify DNA of ghost parasitoids even a month after host death.

2. The coccinellid hosts Cycloneda sanguinea, Eriopis connexa, Harmonia axyridis and Hippodamia convergens were sampled from cole crops in 2017 and 2018 in the Midwest of Brazil, and reared and observed daily for parasitoid emergence. Dead coccinellids were held for 30 days after death before storage and only host carcasses with parasitoid emergence observed were analysed. Species-specific primers were designed for the identification of the parasitoid species that emerged during host rearing: Dinocampus coccinellae, Homalotylus mirabilis, Ho. terminalis, Strongygaster triangulifera and Phalacrotophora sp. The melting temperatures (Tm) of their amplicons were used as positive controls in MCA post-amplification in qPCR.

3. Detection of parasitoid DNA in host carcasses was possible for D. coccinellae, Ho. mirabilis and Phalacrotophora sp. with a limit of detection (LOD) for all the parasitoids <1 pg of DNA, except for Phalacroptopthora sp. (LOD = 1.9 ng). Parasitoids were detected in 31 out of 70 host carcasses (44.3%) with detection ranging from 0% to 71%, depending on the parasitoid species.

4. Synthesis and applications. This work demonstrates that MCA could be used to detect and rapidly identify the DNA of some parasitoid species (e.g. Ho. mirabilis, D. coccinellae and Phalacrotophora sp.) in host carcasses up to a month after parasitoid emergence with high sensitivity and specificity. MCA can be used on any field-collected host and, for some parasitoid species, it should lead to more accurate estimates of the incidence of parasitism. Our approach can be easily adapted and applied to study other parasitoid–host interactions.

qPCR with a Roche 480 (details in published manuscript). Melt curves were constructed using the raw fluorescence data and diffQ in the library MBmca in R (R Core Team 2013) as in the associated R script.

Usage notes are described in Code Book sheet of data file. Software file makes melting curve graphs (-dF/dT), but needs to be modified to accept files in formats different from that described.