Introduction

For many decades, in agriculture and forestry, the main method of plant protection has been the chemical one. Despite the high efficiency, chemicals can negatively affect the environment. A typical example is DDT—banned over 50 years ago—still occurring in agricultural soils1. Furthermore, DDT can accumulate in the bodies of warm-blooded animals2,3 and migrate along food chains4. Modern pesticides such as pyrethroids (cypermethrin)—though more ecologically friendly—still kill bees and freshwater crustaceans5,6. Chitin synthesis inhibitors (lufenuron) accumulate in bottom sediments and can be toxic to benthic organisms such as Chironomus, Gammarus, and others7.

The above-mentioned disadvantages prevent the incorporation of chemical insecticides into production processes related to organic agricultural products, which are currently gaining a market share8 owing to the environmental friendliness of their production. Biological pest control agents are much better suitable for the production of pesticide-free products. Moreover, the complex mechanisms underlying the development of resistance of pests to biological control agents allow these agents to be used against target organisms for a longer period9,10. The most common bioinsecticides are those based on the bacterium Bacillus thuringiensis11. Nevertheless, despite their relatively high specificity and fairly advanced technology of their manufacture, they have some disadvantages. One of the most pronounced drawbacks is the inability to initiate an epizootic that will lead a pest population to self-regulation, which will allow to reduce the frequency of treatments. In this context, viral insecticides are of interest12,13. Among them, polyhedrosis pathogens are the most appealing because they are often found in nature, have high specificity, and are safe for nontarget species14. Viruses of the Baculoviridae family are currently the main sources of viral bioinsecticides11,13. More than 1000 strains of this family are known to infect insects, mainly those from orders Lepidoptera, Diptera, Hymenoptera, and Coleoptera15. Besides, more than 90% of these strains have been isolated from Lepidoptera16: the dominant order of leaf-eating pests. Massive use of baculoviruses for the biocontrol of agricultural and forestry pests17,18,19 has begun to cause the development of resistance in some insect species20,21. The relatively high specificity of viruses from the family Baculoviridae is not only their advantage but also a disadvantage, namely, the complexity of initiating a technological process of mass production of a bioinsecticide against different species of pests. Currently, researchers are trying to overcome this disadvantage by different approaches, for example, by co-occlusion of different viruses in the same occlusion body [for details, see the review by Williams et al.22]. In the meantime the complex replication cycle of these viruses [forming two consecutive morphotypes in different types of host tissues23) imposes substantial restrictions on the rearing of a viral product in a bioreactor; together with a high price of culture media, this situation makes it difficult to organize technological mass production of bioinsecticides based on Baculoviridae representatives.

Our recent study suggests that members of another family of viruses that infect insects have been undeservedly overlooked. This is the family Spinareoviridae and in particular the genus Cypovirus24. Cypovirus (CPV) representatives are often encountered as insect pathogens25,26,27,28. In many cases, however, they induce a chronic infection without causing rapid host death29,30. From the caterpillar Dendrolimus sibiricus, we recently isolated a new cypovirus strain that possesses high efficiency, high productivity, and broad host specificity within the order Lepidoptera and has good potential for the design of a mass production technology24. One of unique features of this strain, DsCPV-1, is its broad host specificity: within the order Lepidoptera, it infects members of families Erebidae, Sphingidae, Pieridae, Noctuidae, and Lasiocampidae24, which are phylogenetically quite distant from each other31. Thus, these features help to distinguish newly discovered strain DsCPV-1 from representatives of the Baculoviridae family. Subsequent studies on DsCPV-1 showed that this strain does possess regulatory effects that negatively affect the next generation of a pest32 or synergistically influence another entomopathogen such as microsporidia33.

Such an important advantage of DsCPV-1 as the wide host range may be a disadvantage in the context of an ecologically friendly strategy of pest control. Moreover, the presence of a consequential mutation in the gene of the A-spike protein of the DsCPV-1 strain24—responsible for virus penetration into host cells—could be the reason for such a broad host range. CPVs mostly infect the Insect class34 and safety toward species other than insects is less likely, although cases of Crustacean infection have been recorded earlier35,36. Whether the recently discovered potential producer of bioinsecticides still has the previously described host range or may infect such important invertebrates as pollinators, predators, or water body inhabitants is still an open question. Moreover, using molecular tools for such a poorly studied genus of viruses as CPV may provide new knowledge about the transmission of CPV via nontarget species.

In accordance with all of the above the purpose of this work was to study ecological safety of using CPV in the field through testing of the strain with the broadest host range (DsCPV-1). Although the idea behind this study is obvious, there are no research articles detailing this issue. Our study was aimed at testing representatives of three types of nontarget organisms: inhabitants of water bodies (where the virus will eventually end up along with surface runoff), pollinators (to assess the safety of DsCPV-1 when used near apiaries), and entomophages/predators (to determine potential suitability of DsCPV-1 for incorporation into integrated pest management [IPM] or to investigate the effect of the virus on invertebrate natural enemies of pests). One of novelties of the current study is that we additionally use a sensitive molecular tool (qPCR) to demonstrate the replication cycle of the virus in nontarget hosts. Finally, this study will be used as guidelines for using the DsCPV-1 strain as a source for mass production of ecologically friendly viral biological agent for lepidopteran pest management in agriculture and forestry.

Results

Bioassay results

The addition of the DsCPV-1 virus into insect containers did not lead to a significant increase in mortality among the tested nontarget aquatic species (Fig. 1).

In some hydrobionts (33% of the tested species), there was a low level of mortality (Fig. 1). At the same time, dynamics of the mortality did not have significant differences between the control and experimental groups of Chironomus sp. (χ2 = 1.036, P = 0.309), A. messeae (χ2 = 0.050, P = 0.823), and C. robusta (χ2 = 0.327, P = 0.568) (Fig. 1, Supplement 1).

In most of the hydrobiont species, there was no mortality in either control or experimental groups. The mortality of bugs P. minutissima and I. cimicoides, bloodsucking mosquito A. aegypti, cladoceran D. magna, G. lacustris, and dragonfly C. lunulatum was 0% in both types of groups (Fig. 1).

Fig. 1
figure 1

Mortality (mean ± SE) at the end of the experiment after exposure (to the DsCPV-1 virus) of Chironomus sp. (5 days after infection initiation); of A. messeae and A. aegypti (8 days after infection initiation); and of D. magna, P. minutissima, I. cimicoides, G. lacustris, C. lunulatum, C. robusta, Apis mellifera and P. maculiventris (day 10 after infection initiation). P. maculiventris (A) DsCPV-1-treated by spraying bugs were fed with Tenebrio molitor larvae; P. maculiventris. (B) bugs not infected with DsCPV-1 were fed virus-infected Manduca sexta larvae.

Full size image

To determine the probability of accumulation and transmission of the virus along trophic chains in the laboratory, a chain was modeled as follows: level 1 consumer → level 2 consumer. Natural hydrobionts G. lacustris and D. magna, exposed to the virus for 3 days, served as level 1 consumers.

The stay of G. lacustris (level 1 consumer) in the water containing the virus did not result in the death of this first participant of the trophic chain (0% mortality). Subsequently, feeding of I. cimicoides (level 2 consumer) on these hydrobionts did not kill the predator either (0% mortality) (Fig. 2).

The stay of D. magna (level 1 consumer) in the water containing the virus also did not cause deaths (0% mortality). Subsequent feeding of the virus-exposed daphnia to P. minutissima (level 2 consumer) did not kill the predator either (0% mortality) (Fig. 2).

Fig. 2
figure 2

Feeding of predators P. minutissima and I. cimicoides on DsCPV-1–treated hydrobionts D. magna and G. lacustris.

Full size image

The susceptibility of terrestrial insects—entomophage P. maculiventris and pollinator A. mellifera—to DsCPV-1 was assessed too. Under our experimental conditions, slight mortality of the tested organisms was noted (Fig. 1). Meanwhile, there were no significant differences in insect mortality between the control and experimental groups of P. maculiventris (χ2 = 0.396, P = 0.529) and A. mellifera (χ2 = 1.037, P = 0.309) (Fig. S1). We also tested the potential effect of virus accumulation in P. maculiventris when the bugs were feed with infected host larvae susceptible to DsCPV-1 (larvae of M. sexta). Again, we did not find significant differences between the treatment and control groups (χ2 = 0.350, P = 0.555) (Fig. 1, S1).

Molecular diagnostic assays

The analysis of four pooled samples of A. aegypti at the beginning of the challenge and at the end of the experiment indicated that only one pooled sample—collected 1 day after water treatment—contained a low signal of DsCPV-1 RNA presence (Table 1).

The assay of cadavers of honeybees who previously fed on honey water inoculated with the virus revealed the presence of viral RNA in half of the tested cadavers. By contrast, control bees that consumed pure honey water and died at same rate did not contain DsCPV-1 (Table 1).

Table 1 Testing samples of different insect species for the presence of DsCPV-1 RNA.
Full size table

Feeding of P. maculiventris with sick larvae of M. sexta (where DsCPV-1 successfully multiplied) expectedly led to a high load of DsCPV-1 in the bugs’ bodies (Table 1). If the bugs fed on sick larvae for the whole study period (10 days), then the virus load in bugs was significantly increased (Fig. 3), implying two scenarios: replication of the virus in M. sexta and/or replication of the virus in P. maculiventris. Our additional experiment revealed that if we excluded infected larvae from the bugs’ diet, then the concentration of viral RNA stopped increasing (Fig. 3). Thus, this means that the source of virus replication was the larvae, not the bugs.

Fig. 3
figure 3

The amount of RNA DsCPV-1 in P. maculiventris at different stadium of infection of infected host M. sexta. (a) continuous feeding by infected M. sexta; (b) feeding by infected M. sexta for 1 day.

Full size image

Discussion

The main criterion for choosing nontarget species for the experimental part of this work was their role in a relevant biocoenosis. Species that are indicators of the environment state are C. robusta, Chironomus sp., and D. magna37,38,39. Species that regulate population density of pest insects are C. lunulatum, P. maculiventris, P. minutissima, and I. cimicoides40,41. The main prey species for other animals, such as fish, are G. lacustris, A. aegypti, and A. messeae42,43. Pollinator species of economic importance include A. mellifera. We believe that the response of these marker species to the exposure to the new pathogenic strain DsCPV-1 makes it possible to relatively objectively assess its virulence toward nontarget species in general and to predict consequences of its application to biocoenoses in particular. We did not test taxa that are more distant from arthropods because it is known that members of the genus Cypovirus do not leave such hosts34,44,45.

As a result of this work, we did not detect significant differences in the mortality of nontarget species between their control and virus-exposed groups. Several tested species are known to be susceptible to other taxa of the genus Cypovirus, e.g., members of the order Diptera (Culicidae, Simulidae, and Chironomidae), subphylum Crustacea (Daphnia)30,35,36,46,47,48, and some insects are known to be susceptible to Baculoviridae: Diptera (Culicidae)49. Our strain DsCPV-1, a representative of the first group of cypoviruses and having a wide enough host range24, proved to be safe for the above species of hydrobionts. According to the results of the molecular diagnostic analysis of DsCPV presence in mosquitos, the probability of the virus being ingested by mosquitoes seems to be very low (only one pooled sample of mosquito larvae out of 4 pooled samples yielded a low signal of DsCPV-1 presence). Despite the absence of the susceptibility of the hydrobionts to DsCPV-1 at present, the risk of future susceptibility due to the reassortment phenomenon which is also inherent in cypoviruses50 cannot be completely ruled out in the case of simultaneous infection of a host with a virulent and nonvirulent strain. This type of coinfections is also characteristic of cypoviruses of the first group28.

It should be noted that the polyhedra of DsCPV-1 are quite large: 1–2 μm24, which means that the virus will settle to the bottom fairly rapidly. Therefore, species residing most of the time within a water layer (G. lacustris, A. messeae, C. lunulatum, I. cimicoides, and P. minutissima) will spatially diverge from the virus, thereby minimizing the impact of the virus on invertebrates. In our assays, however, we also used the typical benthic species (C. robusta and Chironomus sp.) that live most of the time on a muddy bottom. The results of our experiments indicate that for this group, for which the likelihood of contact with the virus is the highest, the virus does not have an adverse effect either.

Our data show that the virus also remains harmless while transmitted along the trophic chain in hydrobiont species: level 2 consumers of hydrobionts did not become infected by the virus when feeding on virus-exposed level 1 consumers. In the literature, we did not find data on the transfer of cypoviruses along trophic chains. Nonetheless, there is information about passive transfer of Baculoviridae strains along a trophic chain with subsequent contamination of an environment. For example, predatory bugs P. maculiventris, while feeding on prey containing such a virus, shed it with feces without being adversely affected by the virus strain themselves51,52. We chose not to detect DsCPV-1 polyhedra in the excrement of predatory hydrobionts and/or the preservation of DsCPV-1’s virulent properties because the main aim of this study was to make sure that there is no negative impact (primarily a decrease in survival) rather than to assess the persistence and spread of the virus in the environment.

When testing DsCPV-1 on several terrestrial insect species, we also demonstrated that it is harmless to pollinators and entomophages. Honeybees demonstrated no differences in the mortality rate even when they drank the honey solution containing high concentrations of DsCPV-1 (lethal for target species), which were intended to create severer conditions in comparison with potential contamination during application in the field. Even under these severe conditions, only half of cadavers contained the viral RNA; together with the mortality data, this finding is most likely a sequence of gut contamination during the feeding without virus replication. For example, an alternative agent of pest control with the same host range, Bacillus thuringiensis, may negatively affect the survival of honey bees53. Thus, the viral insecticide evaluated in the current study is safe for honey bees.

The heaviest virus load was assayed in bugs P. maculiventris. This was done because such predators as P. maculiventris may be used for IPM and be affected not only by a high dose of DsCPV-1 during application in the field but also by a high load of the virus multiplying in infected target larvae. In our both experiments, no significant differences in bug survival were found. Our molecular diagnostic analysis indicates that a huge virus load is present in the bug gut, but it comes from sick larvae and the virus will not replicate in the bug body if we reduce the consumption of the infected material by the bugs.

The results of our study indicate that even strain DsCPV-1 possessing a cross-family host range does not demonstrate cross-order infectivity, as for other CPV strains47,54. Thus, it is unlikely that DsCPV-1 will affect other entomophage taxa, which largely belong to orders Hymenoptera, Hemiptera, Coleoptera, Mesostigmata, and some others. If we compare the influence of DsCPV-1 on nontarget species with effects of most of products available on the market of biological pesticides (products based on B. thuringiensis), then the safety of the viral agent in question may be higher because even with their specificity, B. thuringiensis–based pesticides can have a pronounced effect on nontarget organisms, particularly on bees, terrestrial bugs, Chironomi dae, nontarget Lepidoptera, Culicidae (A. aegypti), and Daphnia4,55,56,57,58.

To sum up, it can be concluded that the application of DsCPV-1 in the field/forest is mostly safe for many invertebrate orders other than Lepidoptera. The only vulnerable group will be rare and endangered species of Lepidoptera, which, judging from the virus’s host range24, may be susceptible to DsCPV-1. On the other hand, application of this biopesticide is still more ecologically friendly than the use of chemical insecticides, where rare insect species are also vulnerable. Current study is the first step of testing the effect of virus on non-target animals and got result is strong background for following testing of virus against vertebrates.

Materials and methods

The virulence of the new strain DsCPV-1 against nontarget organisms was assessed on 11 species of aquatic and terrestrial invertebrates. We modeled the infection of (i) the water body inhabitants that might be infected during insecticide application, (ii) pollinators as the most susceptible component of an ecosystem treated with an insecticide, and (iii) invertebrate predator species widely used for pest management.

Processing of aquatic invertebrates

The choice of hydrobiont species was based on their geographic prevalence, abundance in water bodies, and sensitivity to environmental factors (i.e., so-called ecological indicator species). Aquatic nontarget organisms were represented by both natural and laboratory species. Natural hydrobionts Gammarus lacustris and larvae of blood-sucking mosquitoes Anopheles messeae, of dragonflies Coenagrion lunulatum, Cloeon robusta, Chironomus sp., and of waterbugs Ilyocoris cimicoides and Plea minutissima were collected in July and August 2022 in natural bodies of water in the south of Western Siberia (N 53°43′ E 77°56′). Laboratory species (larvae of the blood-sucking mosquito Aedes aegypti and of the cladoceran Daphnia magna) were provided by the Institute of Systematics and Ecology of Animals (SB RAS) and by analytical service CLATI (Center for Laboratory Analysis and Technical Measurements) in the Siberian Federal District, respectively. Breeding of the laboratory species was implemented in accordance with the conditions necessary for each respective species. The larvae of blood-sucking mosquitoes A. aegypti were kept at 27 °C (± 1 °C) under a photoperiod of 16 h/8 h (light/dark), and the larvae of the cladoceran D. magna were maintained via the standard procedure (ISO 6341:2012).

During the experiments, all the hydrobionts were kept in plastic containers (7.5 × 10 × 4.5 cm) filled with 200 mL of water. Each container housed 20 individuals of D. magna. In cases of A. messeae, A. aegypti, Chironomus sp., or G. lacustris, there were 10 insects per container. C. robusta and P. minutissima were kept at 5 individuals per container, whereas C. lunulatum and I. cimicoides were housed individually. The natural hydrobionts were kept in water from the bodies of water in which they were collected. The laboratory species were collected as follows: A. aegypti in tap water, and D. magna in the Ob River (N 54°59′13″ E 82°42′59″). The water temperature in the containers during the experiments was 25 °C. The blood-sucking mosquito larvae were fed a balanced feed designed for young fish (TetraMin Junior; Tetra GmbH, Munich, Germany), which is used for laboratory rearing of mosquitoes (e.g., Butt et al. 2013; Garrido-Jurado et al. 2016). Water bugs (I. cimicoides and P. minutissima) and dragonflies (C. lunulatum) were fed with the daphnia and the blood-sucking mosquito larvae. Larvae of Chironomus sp., C. robusta, and G. lacustris were fed with algae, plant debris, and small invertebrates that were collected daily in natural bodies of water. Cladocerans D. magna were fed once a day, by the addition of 1.0 mL of a concentrated algal suspension per 100 mL of cultivation water.

Infection was carried out by means of a suspension of DsCPV-1, at 1 mL of the suspension (having a titer of 107 polyhedrons/mL) per container (i.e., per 200 mL of water). This concentration was chosen via testing on the target organisms, many of which have proven to be susceptible to the virus at this concentration (Martemyanov et al. 2023). One milliliter of distilled water (instead of the virus suspension) was added to control containers. In each version of the bioassay, at least 50 individuals of aquatic invertebrates were tested: four replicate (parallel) experiments each with 20 individuals of D. magna (i.e., four containers); 10 containers each with 10 individuals of A. messeae, A. aegypti, Chironomus sp., or G. lacustris; 10 containers each with 5 individuals of C. robusta or P. minutissima; and 70 containers each with one individual of C. lunulatum or I. cimicoides. The work involved larvae of middle-instar (for more reliable species identification and for minimizing the background mortality occurring at younger ages) and 24-hour-old D. magna.

The duration of mortality recording depended on the ontogeny of the invertebrates being analyzed and was 5 days for Chironomus sp.; 8 days for A. messeae and A. aegypti; and 10 days for C. robusta, P. minutissima, I. cimicoides, D. magna, G. lacustris, and C. lunulatum. This duration was justified by the infection process in sensitive hosts24.

We also tested a possible cumulative effect of DsCPV-1 within a trophic chain. In particular, we applied the virus at the aforementioned concentration to G. lacustris and D. magna, which were subsequently fed to predatory hydrobionts I. cimicoides and P. minutissima, respectively. In this experiment, larvae of natural invertebrates I. cimicoides and P. minutissima were kept individually in plastic containers (7.5 × 10 × 4.5 cm) filled with 200 mL of water from their habitats. Hydrobionts characterized by different types of nutrition were provided as food: G. lacustris is omnivorous (served as food for I. cimicoides), and D. magna is a filter feeder (served as food for P. minutissima). Predators were fed 3 times a day. Under the laboratory conditions, G. lacustris and D. magna were fed daily with algae and plant residues that were collected in natural water bodies.

The experiment consisted of two series. The first one was a control. Throughout the experiment, predators of this series were fed with G. lacustris and D. magna unexposed to the virus. The second series was experimental. In it, predators for 3 days were fed with the G. lacustris and D. magna that had been kept in water (200 mL per container) supplemented with 1 mL of the suspension of the DsCPV-1 virus having a titer of 107 virus polyhedra per milliliter. Before the feeding to predators, G. lacustris and D. magna were washed twice in distilled water.

The mortality of the bugs’ larvae was registered daily for 10 days. In each experiment, 70 predator individuals were tested.

Processing of entomophages

The spined soldier bug Podisus maculiventris Say. (Hemiptera: Pentatomidae) is a polyphagous entomophage that feeds on many insect species and prefers soft-shell larval stages. This species has been introduced from North America and is intended for biological protection of plants against pests59,60. Adult individuals of this bug are usually kept in plastic containers equipped with capillary sources of water; once every 2–3 days, depending on consumption, insect prey is added: great wax moths Galleria mellonella, Zophobas morio, or mealworms Tenebrio molitor. During our breeding of the spined soldier bug, air temperature was maintained at 25 ± 1 °C, with air humidity 70–90%, under a 16 h/8 h photoperiod (light/dark). For infection, middle-aged larvae (middle-instar larvae) were housed at 10 per container, 10 containers per treatment. In the experimental group, 10 mL of the virus suspension (108 polyhedrons/mL) was sprayed onto each treatment container. The control containers were treated with an identical amount of distilled water. The concentration of the virus was substantially higher than that used for pollinators/hydrobionts. The reason is that in the case of potentially combined use of predatory bugs with the virus for IPM, the bugs will experience a high direct viral load. Mortality was assessed for 10 days.

Additionally, we simulated the situation in field when bugs have an opportunity to attack sick larvae of target Lepidoptera, where the concentration of a virus might reach 109 polyhedrons per sick larva24. For this purpose, we used Manduca sexta larvae infected by DsCPV-1. The procedure for infecting larvae of M. sexta was described by Martemyanov et al., 2023 24. Bugs P. maculiventris were fed as described in a previous paragraph, with infected/control larvae instead of T. molitor.

Processing of pollinators

A method adopted recently for artificial infection of honey bees61 was employed. Adult worker bees of A. mellifera carnica were obtained from an experimental apiary of the All-Russian Institute of Plant Protection (Russia, St. Petersburg–Pushkin). Insects of different ages (by date of emergence) were collected from the central part of the hives, where younger bees are usually found62. Some collected individuals were analyzed by microscopy to exclude natural infection by pathogens. The collected bees were placed in perforated plastic 0.5 L bottles via a funnel in groups of 20–26 individuals per bottle, each corresponding to a replicate. The insects were fed with 40% sugar syrup poured into a 10 mL glass vial that was closed with a cotton pad and turned upside down, and then the cotton pad was inserted into the bottleneck. To feed the bees with the infectious material, a suspension of DsCPV-1 (adjusted to 0.5 million polyhedra per bee or 107 polyhedra resuspended in 20 µL of water per container) was applied to the surface of the cotton pad that would face the insects. Control insects were treated similarly but distilled water was used instead of the cypovirus suspension. The bottles with bees were kept under natural light for 10 days. In one assay, there were four virus-exposed and four control replicates, and they were kept at 33 °C, which emulates inner conditions of a typical bee hive. To feed the insects, the syrup in the glass vials was replenished upon consumption. Mortality was scored every other day, and the cadavers were removed and dissected for light microscopy to check for possible cypovirus presence.

Molecular diagnostic analysis of DsCPV-1 in the tested species

Diagnostic light microscopy failed to detect massive presence of viral polyhedra, which is a typical symptom of polyhedrosis in a primary host24, although in alternative hosts, the formation of polyhedrons is not informative enough (unpublished data). Therefore, we applied qRT-PCR to be sure in the absence of DsCPV-1 replication in the nontarget hosts63. A molecular tool in the case of occlusion body–forming viruses could be more informative as compared with routinely used light microscopy28. We used A. aegypti, A. mellifera, and P. maculiventris to monitor DsCPV-1 replication in each ecological group. A. aegypti was chosen among all studied hydrobionts as a nontarget host potentially susceptible to chronic CPV infection47.

Because there was no mortality of A. aegypti during virus exposure, we performed additional exposure to DsCPV-1 in order to analyze surviving larvae for potentially chronic infection. We reared two groups of second-instar larvae: one for virus exposure and one as a control. Three 2-liter containers were used per control group and four 2-liter containers per treatment group, at 100 larvae per container. Containers designed for the virus exposure were infected with DsCPV-1 at 108 polyhedrons/container, i.e. in same concentration as the one used for the survival test above. A pooled sample was collected from each container. On the first day after virus exposure, each pooled sample contained 50 larvae per replicate, and on the fifth day after the exposure, each pooled sample contained 20 larvae per replicate. To remove external water, we dried A. aegypti larvae on filter paper, and then the pooled samples were fixed in liquid nitrogen. Until further analysis, the samples were stored at − 80 °C.

In the case of A. mellifera, we observed some background mortality in both control and treated groups, and therefore we used dead bees for the analysis. The cadavers were taken from the pollinator exposure experiment and stored in a dried state until further analysis. For the presence of viral RNA, we checked 5 cadavers from the control group and 14 cadavers from the infected group.

Pooled A. aegypti samples and individual A. mellifera samples were analyzed quantitatively for the presence of DsCPV-1 RNA by the qPCR method described by Akhanaev et al.63, without modifications.

The molecular diagnostic analysis of P. maculiventris was the most difficult. For this diagnostic assay, we used bugs from the heaviest viral load, when animals were fed with sick larvae having increasing virus loads when the larvae showed signs of the disease. There was a difficult with distinguishing the multiplication of the virus in sick caterpillars that are sucked by bugs from true replication of the virus inside the bugs’ gut tissue. To distinguish these two phenomena, we repeated the rearing of the bugs on sick larvae twice: in the first experiment, we fed the bugs with sick larvae for the whole duration of the experiment (10 days) while monitoring the virus load on the fifth and 10th days; in the second experiment, we changed the diet in the treatment group from infected larvae of M. sexta to uninfected ones at 5 days after the beginning of the experiment and fed the bugs until the end of the experiment (the remaining 5 days) with healthy larvae. Thus, if no replication of the virus in the bug gut tissue happened, then we had to see a decrease of the virus load in comparison to the first experiment where the insects continued to suck the sick larvae showing the ongoing DsCPV-1-induced disease.

To infect M. sexta, second-instar larvae were incubated with a suspension of DsCPV-1 polyhedra at 108 polyhedra/mL at a ratio of 0.5 mL/20 caterpillars and then were fed to P. maculiventris starting at 3 days after the inoculation. Thus, each subsequent day, bugs sucked larvae having a higher virus load than on the previous day because of virus replication in M. sexta. Then, on the fifth and 10th day after the exposure, the P. maculiventris individuals were fixed in liquid nitrogen and stored at − 80 °C. P. maculiventris differed from that of A. aegypti and A. mellifera in order to compare Cq values (which are related to the quantity of DsCPV-1 RNA) ​​of P. maculiventris samples on different days after exposure. The sample with the highest RNA concentration from the experimental group was diluted to attain 50 ng cDNA per 25-µl qPCR mixture to avoid inhibition of the qPCR. Other samples were diluted in exactly the same way as the sample with the highest RNA concentration and had a concentration of less than 50 ng cDNA per 25-µl qPCR. In this case, samples that contained identical amounts of target molecules had the same Cq values, and we could compare them. Thus, we analyzed equally diluted RNA samples obtained from P. maculiventris feeding on sick M. sexta larvae to detect the presence/absence of viral replication in P. maculiventris during the experiment.

Statistical analysis

Differences in mortality dynamics were determined in SigmaStat 3.1 (Systat Software Inc.) by the Kaplan–Meier survival analysis followed by the Holm–Šídák logrank test. The data in the graphs are presented as the mean ± standard error (SE). Differences were considered significant at P < 0.05. Data from qPCR were processed by the Mann–Whitney U test (Statistica 6.0).