We recently reported that baculovirus Spodoptera exigua multiple nucleopolyhedrovirus (SeMNPV) triggers positive phototaxis in Spodoptera exigua larvae, leading to death at elevated positions. Dobson et al.  (University of Stirling, Scotland) question our interpretation of the data. Unfortunately, Dobson et al. rely on unwarranted assumptions possibly reflecting a poor understanding of baculovirus–insect pathobiology, make invalid comparisons and fail to take relevant literature into account. Here, we recapitulate the context and interpretation of our experiments and highlight the misinterpretations by Dobson et al.
2. Baculoviruses: masters of manipulation
Baculoviruses are known to manipulate the physiology and behaviour of their larval hosts. Since first described by Hofmann , many studies have demonstrated that baculovirus-infected larvae (i) disperse over larger areas than uninfected larvae (hyperactivity) and (ii) die at elevated positions, typically the tops of trees or plants (tree-top disease) [2–13]. These behavioural manipulations have been described for several baculovirus–larvae systems and may increase viral spread, as virus will rain down plant foliage. Recently, significant progress was made in understanding the mechanism of these manipulations. Viral genes were identified that are required for hyperactivity (ptp1 of Autographa californica MNPV (AcMNPV) and Bombyx mori NPV (BmNPV)) [8,10] and tree-top disease (egt of SeMNPV and Lymantria dispar MNPV (LdMNPV)) [10,14]. Baculovirus-induced hyperactivity and tree-top disease are considered classical examples of parasite-mediated host behaviour.
3. The study by van Houte et al. 
In our study, we identified the cue for tree-top disease, using Spodoptera exigua larvae and SeMNPV as a model system. Three main experiments tested the hypothesis that baculoviruses induce positive phototaxis to trigger climbing behaviour prior to death:
(1) We followed climbing behaviour of baculovirus-infected larvae and found that virus infection causes death at elevated positions (fig. 1a ), which confirms tree-top disease in our experimental system.
(2) Next, we examined whether the height of death was light-dependent. To this end, we exposed infected larvae to light from above or from below, or we kept larvae in the dark. We found that larvae exposed to light from below died at extremely low positions, while larvae lit from above died at high positions. In the dark, larvae died at low positions (fig. 1b ). We concluded that infected larvae moved towards light (positive phototaxis).
(3) Finally, we examined whether climbing behaviour of uninfected larvae is light-dependent. Uninfected S. exigua larvae are known to show climbing behaviour related to moulting, although this behaviour is highly variable [12,13]. We followed climbing behaviour of uninfected larvae under light and dark conditions. We could not detect any meaningful differences in movement patterns between these treatments (fig. 2 ). Hence, climbing behaviour in uninfected larvae is light independent.
Together these data show that baculovirus infection triggers positive phototaxis, causing death at elevated positions.
4. Our response to the commentary by Dobson et al.
Dobson et al.  put forward four arguments for their disagreement with our conclusions.
(1) The association between infection and climbing behaviour is equivocal. This argument surprises us as replicate experiments reproducibly show that baculovirus-infected larvae climb just before they die (figure 1 of this reply; larvae climbing coincides with a drop in survival, highlighted in blue and red); Wilcoxon signed-rank tests: ((figure 1b) interval 75–92 h: Z = −4.478, p < 0.0001; (figure 1c) interval 48–64 h: Z = −2.385, p = 0.017; (figure 1d) interval 65–72 h: Z = −2.897, p = 0.004); Kendall's τ coefficient for correlation between average height of larvae and per cent surviving larvae: ((figure 1b) τ = −0.681, p = 0.033; (figure 1c) τ = −0.802, p = 0.01; (figure 1d) τ = −0.691, p = 0.017). This is in excellent agreement with the literature and in our view the association between infection and climbing is very clear-cut.
(2) Tree-top disease can be explained by two facts: (i) larvae naturally climb and (ii) virus kills, but not instantaneously. This is a fallacy, as diseased larvae generally become lethargic prior to death; climbing prior to death of baculovirus-infected larvae is a striking exception to this rule. In addition, their argument does not explain why larvae in the dark would die at low positions, and ignores a number of relevant studies. First, tree-top disease has been associated with symptoms strikingly different from other types of infections [3,4,6,7]. Moreover, Hoover et al.  identified a viral gene (egt) essential for tree-top disease. We recently confirmed that this gene is required for death at elevated positions in our experimental system .
(3) To measure virus-induced climbing behaviour, behaviour of infected larvae should be directly compared with behaviour of uninfected larvae. We would argue quite the opposite: direct comparisons between infected and uninfected groups are by definition unsound, because baculoviruses strongly suppress development [16–18] (figure 2a). A direct comparison of the behaviour of infected and uninfected larvae at, for example, 70 hours post-infection (hpi) would involve an infected 3rd instar and an uninfected 5th instar. This comparison has two variables (behaviour and development) and is therefore not informative. We therefore used internally controlled experiments in which light is the only variable.
(4) Light-dependent differences exist in climbing behaviour of uninfected larvae. This argument is based on an invalid comparison. Dobson et al. compare larval height under dark and light conditions at different time points, which is biologically meaningless; phototaxis should be measured by comparing larval height under dark and light conditions at the same time point (figure 2b,c). A correct analysis of phototaxis in uninfected larvae does not support the argument by Dobson et al.: climbing behaviour is qualitatively identical, with both moults and hence climbing peaks related to moulting (from 3rd to 4th instar and from 4th to 5th instar), occurring at approximately the same time and height under dark and light conditions. If climbing of uninfected larvae would have been the result of positive phototaxis, we should have seen greatly reduced climbing peaks for uninfected larvae kept under dark conditions. This is not the case, and the large differences in height of death of infected larvae can therefore not be explained by pre-existing positive phototaxis in uninfected larvae, but instead can only be explained by virus-induced positive phototaxis. In this context, it is interesting to note that—unlike 3rd and 4th instar larvae—1st and 2nd instar larvae do display positive phototaxis , and these existing pathways may well be hijacked by the virus to induce positive phototaxis in infected 3rd instar larvae.
Finally, Dobson et al. argue that the parasite may cause death at an optimal time post-infection. Virulence (which determines time to death) is a key parasite life-history trait, and its importance is extremely well documented (e.g. [20–23]). We do not expect this to be different in the context of parasite-mediated host manipulation, especially considering how well-timed the observed manipulation is (i.e. larvae climb shortly before they die; figure 1). Indeed, recent data from our group confirm that an altered time to death impacts the tree-top disease induced by baculoviruses .
5. Conclusion and outlook
In summary, the commentary by Dobson et al. fails to convince us and does not invalidate our main conclusion that light is the cue for ‘tree-top’ disease. Their arguments are based on invalid assumptions and comparisons that are borne neither from our experimental design nor from the biology of our experimental system. That said, we consider the commentary by Dobson et al. as a sign of keen interest in the triggers of host manipulation and we invite them to experimentally invalidate our conclusion that light is the cue for an altered host response resulting from SeMNPV infection. With the identification of a virus gene necessary for behavioural manipulation , we will focus our future research on understanding the mechanism behind baculovirus-induced behavioural manipulation and the adaptive significance of this fascinating behavioural manipulation.
Data are available from the dryad repository: doi:10.5061/dryad.2fk77.
We declare we have no competing interests.
We received no funding for this study.
The accompanying comment can be viewed at http://dx.doi.org/10.1098/rsbl.2015.0132.
- Received July 21, 2015.
- Accepted September 15, 2015.
- © 2015 The Author(s)
Published by the Royal Society. All rights reserved.