Plant-derived feeding deterrents have attracted increasing attention in integrated pest management [1] . Despite this agricultural importance, there have been only a limited number of applicable products [2] , almost none avai- lable on highly polyphagous species like Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae). It has been demonstrated that specialist herbivores showed greater sensitivity and lower habituation to plant-derived feeding deterrents than generalists [3] , and thus it can be expected that an effective feeding deterrent targeting to a polyphagous species would also control the co-occurring oligophagous or monophagous ones [4] .

H. armigera is one of the most important economic insect pests in many parts of the world. The highly polyphagous larva has been recorded as a major pest of maize, sorghum, tomato, lucerne, tobacco, cotton, and cowpea. Extensive insecticide spraying resulted in low yields and high control costs, and the ever-increasing resistance has led to a renewed interest in developing alternatives to insecticidal control [5] . With this respect, several deterrent plants or their extracts have been reported for H. armigera [6] - [9] . However, common spices with strong smells or heavy tastes would provide a promising feeding deterrent source, but have not received enough attention.

We previously found the ethanol extract of Zanthoxylum bungeanum, a numb-taste spice, strongly inhibited H. armigera larvae feeding [10] , and thus it was an impetus to compare more spices. The first phase of this study involves the laboratory screening and comparison of the potential relative feeding deterrence of various pungent or hot-tasting spices so as to identify a few promising ones for subsequent detailed investigations.

The mean feeding amounts and FDI values of each group are summarized in Table 2. Generally, the overall feeding amount difference in each group is significant, as shown in Table 2, and the reference sample always evoked significant feeding deterrence. In Group 1, the feeding amounts of leaf discs coated with the extracts of P. longum, Z. bungeanum, and M. fragrans were significant lower than that of the control, and P. longum is the strongest material with a FDI value 88.76 and mean feeding amount only 3.95 mm². In Group 2, E. globulus extract exhibited significant feeding deterrent activity but not stronger than the reference. In Group 3, the extract of A. dahurica produced the strongest activity with a FDI value as high as 98.03, the mean leaf area consumed was only 1.00 mm², followed by I. verum extract being equal to the reference. In Group 4, P. nigrum (white pepper) evoked the maximum FDI value of 97.49, and the mean leaf area consumed was only 0.88 mm². In Group 5, P. nigrum (black pepper) extract had the strongest feeding deterrence activity with a FDI value of 93.86, and the mean leaf area consumed was only 1.05 mm², followed by F. vulgare, Z. bungeanum, and C. reticulate. However, in multiple-choice bioassay, a sample exhibiting “significant” effect may be caused by its members, as shown by the FDI values of the reference in each group (ranging from 42.34% to 55.50%), we believe that regrouping may resolve this problem.

Groups 6, 7, and 8 was on the basis of the above results. In the second round, the extracts of I. verum, C. reticulata, M. fragrans, E. globulus, and F. vulgare were excluded from the active extract list, because their mean

^aMean feeding amounts followed by “^*” and “^**” within each group are significantly different from the corresponding control at P = 0.05 and P = 0.01 levels, Friedman test with Bonferroni-Dunn correction.

percent feeding amounts had not significant difference compared to their corresponding controls. Taken together, the three materials from Piperaceae [P. nigrum (black), P. nigrum (white), and P. longum] and A. dahurica had the strongest and almost equally feeding deterrence to H. armigera larvae.

The use of insecticides is a double-bladed sword in crop protection. Although they no doubt play an important role in protecting the crops from damage, adverse effect on natural enemies also can not be ignored. Feeding deterrents may be an alternative to resolute this technical contradiction since they may protect crops from damage meanwhile offered the natural enemies with “non-toxic” prey. To date, the most widely used and most successful feeding deterrent is azadirachtin, whose anti-feeding effect is highly variable among pest species, and even those species initially deterred are often capable of rapid desensitization [2] , additionally, its noxious bitter taste further limited its application in some frequently harvested crops such as vegetables. From a prospect of practice, the edible spices would be a promising resource of feeding deterrents.

With this regard, we tested the feeding deterrence of some common edible spices against H. armigera. These includes four species from Apiaceae, three species each from Rutaceae and Zingiberaceae, two species from Lauraceae, and one species each from Brassicaceae, Illiciaceae, Myristicaceae, Myrtaceae, Solanaceae, and Valerianaceae. Among these, all the Piper materials exhibited strongest and equal feeding deterrence. Similarly, Piper cenocladum and P. imperiale were deterrent to Spodoptera frugiperda, another noctuid generalist [4] , and other herbivores such as two Hymenoptera species (Paraponera clavata and Atta cephalotes) and two Orthoptera species (Microtylopteryx hebardi and Homeomastax robertsi) [16] [17] . However, to the best of our knowledge, this is the first report on the feeding deterrence activity of P. nigrum and P. longum towards H. armigera. This study and the previous reports suggest that the potent feeding deterrent activity may be a common characteristic at genus level. In contrast, Z. bungeanum always evoked fairly strong feeding deterrence within each group, while Z. schinifolium did not, although they belong to the same genus, indicating that even phylogenetically closely related plants did not necessarily share common defense traits.

The phytochemistry of the family Piperaceae and the genus Piper is well documented [18] [19] , many species contain insecticidal secondary metabolites, providing a broad array of defenses against arthropod herbivores, the predominant ingredient is piperine [16] . Furthermore, the extract of another related species, Piper guineense, could significantly inhibit the hatching of the newly-deposited eggs of Maruca vitrata and Clavigralla tomentosicollis [20] , and suggesting that the action mode of Piper species is more complex than had been realized, although the active compounds inhibiting egg hatching have not been characterized yet. A. dahurica was the only one plant species exhibiting potent feeding deterrent activity among the four Apiaceae plants, which mainly contained a variety of furanocoumarin derivates [21] [22] . Unfortunately, these compounds are not available in our laboratory presently.

A potential feeding deterrent may act via the following mechanisms: 1) olfactory repellent effect, in this case, the larvae will reject the treated feeding substrate without contacting, 2) true taste deterrence, then the larvae will cause some small biting points on the substrate, 3) physical obstacle via oral mechanoreceptors, and/or 4) toxicity via post-ingestive response, if this is true, the larvae will consume relatively much food before they reject them [23] . In our bioassay, the first and the second possible mechanisms could not be discriminate yet, but the fourth mechanism cannot explain our results since all of the test subjects were living after bioassay.

In conclusion, only four materials were finally determined as the effective feeding deterrent resources of H. armigera larvae, except for Z. bungeanum, which was determined in a previous study. The isolation and purification of crude extract of Piper species and A. dahurica, as well as their long-term effect and possible mechanism involved in feeding deterrence, deserve to be explored further.

This work was granted by Natural Science Foundation of China (31471772) and The National Special Research Fund for Non-Profit Sector (Agriculture) (201203036). Special thanks are given to the following undergraduate trainees, Ruihong Wang, Li Zhuang, Ruilin Wang, Liwei Han, Jinlei Li.